fix: check max depth for every depth a fragment is spread at

[mwhooker]

validateMaxDepth guards against fragment cycles with a single visited set for the whole operation. Once a fragment has been checked, every later spread of it is skipped, including spreads at a greater depth, where its fields are deeper. A query can exceed MaxDepth by spreading a fragment shallowly before spreading it deeply, or reach any depth with a chain of fragments that are each spread once near the top first.

This keys visited by fragment and depth instead. A spread at a depth that has already been checked is still skipped, so cycles still terminate, and depth never exceeds maxDepth+1 because fields beyond the limit are not descended into, so each fragment is walked at most maxDepth+1 times.

The existing spreadAtDifferentDepths case covered this scenario but passed only because its query also failed type validation (PossibleFragmentSpreadsRule, FieldsOnCorrectTypeRule); no MaxDepthExceeded error was produced. It now uses a valid query and asserts MaxDepthExceeded, and fragmentChainSpreadShallowFirst covers the chain. Both fail on main.

[Copilot]

Copilot review overview

🟢 Approval recommended

The depth-aware traversal is correct, bounded, and adequately covered by regression tests.

Review effort: Balanced
Findings: None

What changed in this PR

Fixes max-depth validation for fragments reused at different nesting depths.

Changes:

• Tracks visited fragments by definition and depth.
• Adds valid regression tests for direct and chained fragment spreads.

File	Description
internal/​validation/​validation.go	Corrects fragment depth traversal and cycle handling.
internal/​validation/​validate_max_depth_test.go	Adds targeted regression coverage.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.