feat(asap-tools): Separate ClickHouse ingest and query CPU/memory monitoring in experiment_run_clickhouse #521

Summary

Baseline ClickHouse experiments do not cleanly separate resource usage by phase: monitoring typically wraps only the query client, so **bulk ingest CPU/memory is missing **. For fair comparison against SketchDB, we need distinct monitoring for:

1. Bulk ingest (JSON load into ClickHouse)
2. Query (client workload against ClickHouse)

SketchDB already gets a continuous precompute+query monitor with per-thread attribution. Baseline should expose a comparable ingest-only signal via a dedicated monitor output.