fix(voice/consistency): vary stock formulas, vendor-balance orient.py, unify the loop

- Vary 11 instances of two boilerplate openers ("A generic X course…" /
  "Strip away X…") across 10 modules so they read as distinct, concrete prose;
  kept the few deliberate, voice-distinct uses; locked exemplars untouched.
- M23 orient.py: detect a vendor-balanced set of AI-instruction filenames
  (AGENTS.md/CLAUDE.md/GEMINI.md/.cursorrules/.cursor/rules/copilot-instructions)
  instead of singling out one vendor. Still runs.
- Render the collaboration loop consistently as seven stations
  (issue->branch->implementation->PR->review->merge->closed) in M25/M26 to match
  M11 and the syllabus.

Closes #48
Closes #49
Closes #51

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TfzV5QvtPDz8LJS3Pu5VLT

modules/23-working-with-existing-codebases/README.md

@@ -144,10 +144,10 @@ in unfamiliar code," they encode *exactly* what careful means, as steps the AI f
 
 ## The AI angle
 
-A generic "onboarding to a legacy codebase" guide would tell a human to read the README and ask a
-senior dev. What's specific here is that **the AI is both the thing reading the codebase and the
-thing most likely to confidently misread it** — and the bigger the repo, the wider that gap between
-"sounds authoritative" and "is correct."
+Onboard a human to a legacy codebase and the advice is familiar: read the README, ask a senior dev.
+What's specific here is that **the AI is both the thing reading the codebase and the thing most
+likely to confidently misread it** — and the bigger the repo, the wider that gap between "sounds
+authoritative" and "is correct."
 
 So the AI-specific discipline is verification, not exploration. The model is genuinely excellent at
 the grunt work of orientation — reading a hundred files, summarizing structure, tracing a call path —