Refactor TokenStreamTranscriber duplication

[ysdede]

Extracted shared logic for result merging, state updates, and callbacks into _handleProcessingResult to fix code duplication in TokenStreamTranscriber.ts. Added unit tests in src/lib/transcription/TokenStreamTranscriber.test.ts. Verified with new tests.

---

PR created automatically by Jules for task 18133579909554393049 started by @ysdede

[ysdede]

Review Summary (Post r1→master merge)

Status: ✅ MERGEABLE

Changes: Refactors TokenStreamTranscriber to eliminate code duplication between processChunk() (audio path) and processChunkWithFeatures() (precomputed mel path). Extracts shared post-processing logic into a new _handleProcessingResult() private method.

Assessment: Clean DRY refactor. Both methods had ~30 lines of identical merge/format/callback logic that is now consolidated. The extracted method handles:

• LCS merger processing
• State updates (timestamp, chunk count)
• Text formatting via tokenizer
• Callback notifications
• Building the return result

No behavioral changes, just structural improvement. Makes future pipeline additions easier.

Recommendation: ✅ Merge — Good code quality improvement, reduces maintenance burden

[google-labs-jules]

Review Summary (Post r1→master merge)

Status: ✅ MERGEABLE

Changes: Refactors TokenStreamTranscriber to eliminate code duplication between processChunk() (audio path) and processChunkWithFeatures() (precomputed mel path). Extracts shared post-processing logic into a new _handleProcessingResult() private method.

Assessment: Clean DRY refactor. Both methods had ~30 lines of identical merge/format/callback logic that is now consolidated. The extracted method handles:

• LCS merger processing
• State updates (timestamp, chunk count)
• Text formatting via tokenizer
• Callback notifications
• Building the return result

No behavioral changes, just structural improvement. Makes future pipeline additions easier.

Recommendation: ✅ Merge — Good code quality improvement, reduces maintenance burden

Acknowledged. Thank you for the review.