[BENCH-835] Fix STT/TTS normalized validator false negatives

[kdmelon]

Summary

Failed TTS requests with no audio filename and no TTFA measurement now reconcile only when an exact live-written provider-failure observation exists, including the canonical null timing artifact and its evaluation input. These cases remain ineligible for backfill.

STT observations whose metrics all fail with distinct valid errors now reconcile as succeeded observations with failed evaluations. Shared provider errors and ambiguous or malformed records retain their existing handling. Rollup comparison and read cutover behavior are unchanged.

Closes BENCH-835.

Testing

• Full tests/unit/test_normalized_storage_backfill.py: 94 passed, including PostgreSQL-backed cases.
• Changed-file Ruff check/format, strict mypy, and git diff --check: passed.
• Local PostgreSQL 16.10 smoke used the real RunWriter and dual_write, then ran the migration CLI twice with database writes disabled. Both failure shapes reconciled with zero mismatches, identical reports, and unchanged table snapshots. Temporary servers were stopped.

Deployment and production cutover validation remain separate rollout steps.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Advanced

Run ID: 7f26c081-9e0a-4a78-a2b8-f58820ba350e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}