fix(tts): include status, attempts, and last HTTP code in LucyLab polling timeout

[nguyentran4896]

Summary

• Problem: When a LucyLab TTS polling job times out, the original error message was generic: LucyLab export <id> polling timeout after <ms>ms. Users had no way to distinguish slow processing, server-side failure, auth problems (401/403), or network issues.
• Fix: The polling loop now tracks lastStatus, pollAttempts, elapsed seconds, and the last HTTP status/response snippet. On timeout, the error includes all of this context.
• Tests added: Two new test cases verify (a) the error message for a stuck pending job contains the job ID, last status, and attempt count; (b) when all polls return 401, the error says "no successful poll" and includes the HTTP 401 status.

Before / After

Before:

Error: LucyLab export exp-abc polling timeout after 60000ms

After (stuck pending):

Error: LucyLab export exp-abc polling timeout after 60s (60000ms limit). Attempts=12. last status="pending", HTTP=ok. Check: (a) API key valid? (b) job status on LucyLab dashboard? (c) network connectivity?

After (auth failure):

Error: LucyLab export exp-abc polling timeout after 60s (60000ms limit). Attempts=12. no successful poll, last HTTP=401, response="{"error":"Unauthorized"}". Check: (a) API key valid? (b) job status on LucyLab dashboard? (c) network connectivity?

Test plan

• npm run typecheck passes (0 errors)
• npm test passes (46 tests, 9 test files)
• New tests cover stuck-pending timeout (job id + status + attempts in message)
• New tests cover all-polls-fail-401 timeout (HTTP 401 + "no successful poll" in message)

🤖 Generated with Claude Code