Extract grok-ozempic#42 once review-signal dedupe + language_by_pr land

[rmems]

Goal

Add rmems/grok-ozempic#42 (feat: export Grok-1 embedding pickle → .npy) to grok-ozempic-v0 as an 8th record.

It was deliberately deferred, not rejected, during the #11 extraction (#17).

Why it is worth having

• 51 signals survive the bot filter — 4th highest in the repo
• 17 commits, +996/−6, ships a real test file (scripts/test_export_grok1_embedding_npy.py, +261)
• Would be the dataset's first Python record; everything extracted so far is Rust
• Advances grok-ozempic#37 / RM-189

Blocked by

• normalize: dedupe review_signals and deprioritize 'Addressed in <sha>' acks #18 — signals currently collapse to 8 byte-identical Addressed in <sha>: … acks, so the record carries no review→patch shape
• card: support language_by_pr override in normalize.language_for #19 — the PR is Python but grok-ozempic-v0.json declares "language": "Rust", and language_for has no per-PR override

Tasks

• Confirm normalize: dedupe review_signals and deprioritize 'Addressed in <sha>' acks #18 and card: support language_by_pr override in normalize.language_for #19 have landed
• collect_pr_records.py --repo rmems/grok-ozempic --pr 42 --out-dir datasets/raw/grok-ozempic
• Add 42 to the card: pr_numbers, pr_count 7→8, training_use_buckets (review-to-patch), domain_by_pr (ml-infra), language_by_pr (Python)
• Rebuild JSONL + regenerate manifest (sha256/bytes/record_count all change)
• Verify ≥5 unique review signals and language: Python on the new record
• Update the shortlist and measured-yield tables in docs/source-repos.md

Follows #11.

[linear-code]

RM-227

[coderabbitai]

🔗 Related PRs

#9 - docs(source-repos): add corinth-canal v0 extraction shortlist (#4) [merged]
#12 - feat: read-only PR collector + corinth-canal-v0 trajectory JSONL [merged]
rmems/grok-ozempic#42 - feat: export Grok-1 embedding pickle → .npy (#37 / RM-189) [merged]
#17 - feat: grok-ozempic-v0 trajectories (revised 7-PR shortlist) [merged]
rmems/grok-ozempic#79 - fix: harden #75 secondary evidence validation (PR #76 bot review) [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

✅ Successfully generated implementation plan.

💡 Use @coderabbitai <feedback> to replan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!