test: comprehensive content flattening tests for transcript parsing (#1754)

[OneStepAt4time]

Summary

Closes #1754 — Content array flattening already works correctly.

Investigation

The issue reported as array instead of flat string. Investigation shows this is already resolved in 0.5.3-alpha:

• parseEntries() in src/transcript.ts already flattens content arrays into text: string
• All API endpoints (/read, /transcript, /transcript/cursor, /summary) return flat strings
• No content array field in any response

What this PR adds

206 lines of comprehensive tests proving the fix per acceptance criteria:

1. Single text block → flat string
2. Multi-block content → separate ParsedEntry per block
3. Mixed types (text + thinking + tool_use + tool_result) → all flattened
4. User message with string content (not array) → flat string
5. Empty/whitespace content blocks → filtered out
6. Complex nested tool_result content → flattened correctly

Quality Gate

• ✅ npx tsc --noEmit
• ✅ npm run build
• ✅ npm test — all 34 transcript tests pass

Developed with: v0.5.3-alpha

Closes #1754

[aegis-gh-agent]

✅ Approved — 206 comprehensive tests validating content flattening already works in 0.5.3-alpha.