Refresh Challenge Coach podcast transcripts after PR #202 / #203 (autograding + registration migrations)

[accesswatch]

Summary

After PR #202 (autograding migration) and PR #203 (registration simplification) land, several Challenge Coach podcast transcripts under podcasts/transcripts/challenges/ will describe behavior that no longer matches the workshop.

These transcripts are paired with already-rendered audio (75 episodes total across chapters, challenges, and appendices), so this is a re-record pass, not a quick text fix.

What is now stale

Autograding-related drift (from PR #202)

Transcripts reference 'the autograder' as a Classroom test runner and describe a points-table comment with '10/10 or 0/15' scores. After PR #202, every check is a separate GitHub Actions workflow that posts a single named comment per challenge (## Challenge N: ...) with deterministic English feedback -- no points, no pass/fail table.

Known affected transcripts:

• cc-07-survive-a-merge-conflict-segments.json
• cc-10-go-local-segments.json
• cc-14-design-an-issue-template-segments.json
• cc-16-build-your-own-agent-segments.json -- multiple long segments about the autograder comment shape
• cc-bonus-a-improve-agent-segments.json -- same autograder language

Full surfacing query: grep -ri 'autograding-day\|points/15\|pass/fail status\|autograder comment with a table' podcasts/transcripts/.

Registration-related drift (from PR #203)

If any transcript references org invites, CLASSROOM_ORG_ADMIN_TOKEN, or 'you'll get an invitation to join the org', that language is now incorrect -- the welcome comment posts the two assignment URLs directly.

Suggested next steps

1. Run grep -ri 'autograder\|autograding\|org invite\|CLASSROOM_ORG' podcasts/transcripts/ to enumerate every affected segment.
2. For each affected episode, regenerate the draft transcript: npm run generate:podcast-transcripts.
3. Listen-QA the regenerated draft, hand-tune the host banter, then re-render audio: npm run build:podcast-audio:challenges.
4. Re-tag MP3 metadata: npm run podcast:metadata:write.
5. Validate inventory and feed: npm run podcast:inventory:check and npm run validate:podcast-feed.

Out of scope

• Re-recording chapter/appendix episodes unless grep shows they reference deprecated behavior.
• Audio hosting / RSS feed (tracked separately in Regenerate podcast audio files and fix RSS feed #182).

Why not block PR #202 / #203 on this

The text changes shipped in PR #202 and #203 are the source-of-truth fix. Audio is a downstream artifact with its own QA cost. Async learners are best served by getting the corrected written docs and html/ pages live promptly, then refreshing audio in a focused pass.

Acceptance criteria

• All challenge transcripts that previously named the Classroom autograder or org-invite flow describe the workflow-based check + welcome-comment-URL flow instead.
• Affected MP3s re-rendered and re-tagged.
• npm run podcast:inventory:check and npm run validate:podcast-feed pass.
• Manifest counts unchanged (still 75 episodes).