refactor(planner): move Pass 1 helpers out of the legacy search; drop show_post_asap_ir, erp and empirical_comparison

[zzylol]

Why

This is the first step in removing the legacy replacement search. The rule is that a legacy piece is deleted only once the new architecture implements it, and the group that reimplements a capability deletes the legacy piece. This PR deletes only the pieces that are already replaced (show_post_asap_ir → stage_pipeline) or that the user decided to drop (erp, empirical_comparison: Assumption 5's deployment-owned empirical models cover them, and sketch-bench keeps its own copy). It also moves the helpers Pass 1 shares with the legacy search out of replacement.rs, so that file can be deleted later without touching Pass 1.

What stays, and which group or issue removes it:

• recurrence, SharedSubDAGStrategy: G4
• accuracy::reconciliation: G5
• HLL confidence wrapper: Wire the HLL confidence model into Stage 1 sizing and plan selection #678
• function_rules, accuracy::composition, exact_composition: Compose accuracy guarantees through arithmetic and comparisons over approximate values #669, with an accuracy::allocation follow-up
• pane_sharing: G6/G10
• grouping: G7
• cost_model capability hooks: G9
• empirical_cost, empirical_resources, physical_plan_cost_model: Calibrate plan selection's costs from measured sketch costs #679
• rollup, topk_reuse, nested-sum: Pass 2 reuse rules: rollup, top-k reuse, nested sum merge #680
• population adapter: Offer maintained-population top-k candidates in Pass 1 #681
• dag_export, asap_types::{dag_export, cost}: the early viewer rebuild
• legacy core (replacement search, rewrite, explanation, propagate, PlanningModels.cost, the legacy integration tests and evidence types): last

What

• New asap_aware_mapping::realization holds summary_candidates, accuracy_target, accuracy_budget, default_size_params, DEFAULT_DELTA, column_ref, summarised_column and summarised_input. The bodies moved unchanged from replacement.rs. The one difference is that summarised_column is now pub(crate), because the legacy search still calls it. Pass 1 (per_computation_candidates), replacement.rs and the other callers import from the new module. RealizationError also moves to realization (re-exported from replacement), so the new module does not depend on the legacy search.
• Deleted the show_post_asap_ir bin, which stage_pipeline replaces. Renamed show_pre_asap_ir to show_logical_dag. Bins are auto-discovered, and no CI or tests referenced either name.
• Renamed the frontend-common resolver's fallback schema to usage_derived_schema.
• Deleted erp.rs and empirical_comparison.rs, plus empirical_cost::snapshot_prepare_cpu (only empirical_comparison used it) and its unit test.
• No Markdown changes.

Key code interfaces added / modified

crates/asap-aware-mapping/src/realization.rs (new), replacement.rs, lib.rs

// realization.rs: Pass 1 helpers moved unchanged out of replacement.rs
/// Confidence δ assumed when the target carries only an ε; (ε, δ)-sized
/// sketches such as CMS need one.
pub const DEFAULT_DELTA: f64 = 0.01;
/// The sketch kinds that can serve an intent, most-preferred first; Pass 1
/// reads it to list an intent's summary candidates.
pub fn summary_candidates(intent: &AggIntent) -> &'static [SketchAlgorithm];
/// The accuracy target on an approximate-capable intent (quantile,
/// cardinality, count, top-k, ...), or `None` for every other intent.
pub fn accuracy_target(intent: &AggIntent) -> Option<&AccuracyTarget>;
/// Resolves an accuracy target into the `(eps, delta)` budget sketch sizing
/// needs, filling δ with `DEFAULT_DELTA` when only ε is given.
pub fn accuracy_budget(accuracy: &AccuracyTarget) -> (f64, f64);
/// The built-in sketch sizing for `(eps, delta)`: the smallest parameters
/// meeting the bound, clamped to the family's range.
pub fn default_size_params(kind: SketchAlgorithm, intent: &AggIntent, eps: f64, delta: f64) -> SketchParams;
/// The one column a single-column summary reads, or the PromQL sample value
/// when the intent reads none.
pub(crate) fn summarised_column(intent: &AggIntent, child_schema: &Schema) -> ColumnRef; // was private
/// A schema field as a named or table-qualified column reference.
pub(crate) fn column_ref(column: &Field) -> ColumnRef;
/// What the summary consumes per row: one column, or the whole tuple for a
/// multi-column intent; a tuple column outside the schema is an error.
pub(crate) fn summarised_input(intent: &AggIntent, child_schema: &Schema) -> Result<SummaryInputExpr, RealizationError>;

/// Why building a candidate's plan failed.
pub enum RealizationError { // moved from replacement.rs
    /// Schema derivation failed while binding a sub-DAG.
    Schema(SchemaDerivationError),
    /// The candidate cannot meet its accuracy target (no sound propagation
    /// rule, or the target is missed); it is rejected, not built.
    Accuracy(AccuracyError),
    /// The summary family's realization rule cannot represent the complete
    /// input (an unsupported shape).
    PhysicalRealization(&'static str),
    /// A constructed plan violates the update/evaluation phase contract;
    /// detected at construction.
    ExecutionDataState(ExecutionDataStateError),
    /// An exact operator's output schema could not be derived over its child,
    /// which carries summary state it can't read.
    ExactOperationSchema(ExactOperationSchemaError),
}

// replacement.rs
/// Callers that import the error from `replacement` keep compiling.
pub use crate::realization::RealizationError; // re-export keeps the old path

// lib.rs
/// Pass 1's candidate and sizing helpers.
pub mod realization;                         // new
/// Crate-root path for listing an intent's sketch candidates.
pub use realization::summary_candidates;     // was replacement::summary_candidates
// removed: pub mod erp;
// removed: pub mod empirical_comparison;

crates/asap-aware-mapping/src/erp.rs, empirical_comparison.rs, empirical_cost.rs (deleted files linked at the parent commit)

// erp.rs, removed (deployment-owned empirical models cover it)
/// A versioned file of error-resource profile records.
// removed: pub struct ErpArtifact { .. }
/// One benchmarked sketch configuration: parameters, data distribution, observed errors, resources.
// removed: pub struct ErpRecord { .. }
/// A record's memory and update/merge/query CPU costs.
// removed: pub struct ErpResourceProfile { .. }
/// Inputs for picking the cheapest record under an empirical error bound.
// removed: pub struct ErpSelectionRequest { .. }
/// Unsupported schema version, invalid artifact, or no record meets the error bound.
// removed: pub enum ErpError { .. }
// removed: /* ErpDataShape, ErpShapeFit, ErpMultiFitSelectionRequest, ErpNearestSelectionRequest, ... */

// empirical_comparison.rs, removed for the same reason
/// Compared fixed-snapshot CPU costs of sketch and exact alternatives and recommended one.
// removed: pub fn recommend_offline(evidence: &OfflineComparisonEvidence, request: &OfflineComparisonRequest) -> Result<OfflineRecommendation, String>;
/// Whether a sketch's parameters are componentwise at least a minimum.
// removed: pub fn parameters_at_least(candidate: &SketchParams, minimum: &SketchParams) -> bool;
/// Query, accuracy requirement, workload and weights for `recommend_offline`.
// removed: pub struct OfflineComparisonRequest { .. }
/// The selected option, the exact baseline, every candidate and the estimated savings.
// removed: pub struct OfflineRecommendation { .. }

// empirical_cost.rs
/// A measurement row's snapshot preparation CPU, if known.
// removed: pub(crate) fn snapshot_prepare_cpu(row: &OfflineMeasurement) -> Option<f64>; // only empirical_comparison used it

crates/devtools/src/bin/show_logical_dag.rs (renamed from show_pre_asap_ir.rs), show_post_asap_ir.rs (removed)

// Lowers a batch of SQL/PromQL queries to pre-ASAP IR and prints it; picks no sketch.
// bin show_logical_dag: renamed from show_pre_asap_ir
// Printed the post-ASAP IR; use stage_pipeline for the planner stages instead.
// bin show_post_asap_ir: removed; stage_pipeline replaces it

Usage:

use asap_aware_mapping::realization::{accuracy_budget, accuracy_target, default_size_params, summary_candidates};

if let Some(target) = accuracy_target(intent) {
    let (eps, delta) = accuracy_budget(target);
    for kind in summary_candidates(intent) {
        let params = default_size_params(kind.clone(), intent, eps, delta);
    }
}

Verification

• git grep -w finds no remaining reference to any deleted public item or to either old bin name outside Markdown.
• cargo fmt --all -- --check and cargo clippy --workspace --all-targets -- -D warnings: clean.
• cargo nextest run --workspace: 1609 passed, 50 skipped. All 23 removed tests were in deleted code: 10 in erp, 9 in empirical_comparison, 3 in show_post_asap_ir, and 1 for snapshot_prepare_cpu.
• stage_pipeline --example planner-layering-1 output is byte-identical to tools/dag-viewer/examples/planner-layering-example1.json.
• tools/dag-viewer/test_render.py: 29 OK. tools/verify_metricsql_external_consumer.sh: OK.

Human review

• The moved helpers and RealizationError are unchanged
• Deleting erp/empirical_comparison (and snapshot_prepare_cpu) matches the decision
• The ownership list above for what stays is correct