## Summary

- stop generating lexer and parser snapshot tests
- remove 488 lexer and parser snapshots and their unused formatting
helpers
- update compiler test documentation

## Testing

- cargo test -p baml_tests --lib (1428 passed, 6 ignored)

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **Documentation**
* Updated contributor testing guidance to focus on compiler IR and
diagnostics, removing references to lexer/parser snapshot outputs.
  * Adjusted the “What Gets Tested” overview accordingly.
  * Fixed formatting for benchmark command blocks.
* **Tests**
* Stopped generating lexer and parser snapshot tests, while keeping
later-stage coverage (HIR/THIR, diagnostics, code generation, and
related tests) intact.
* Simplified the test harness to rely on the generated test suite
content directly.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->