v0.5.1 — Deep Research Transient Error Handling

🐛 Bug Fix: Issue #98

Deep research (--mode deep) intermittently failed with a generic "no confirmation from API" message when Google returned a transient error. The structured error payload (e.g., DeepResearchErrorDetail code 3) was silently swallowed because _extract_rpc_result() only checked for auth errors (code 16).

Now instead of a cryptic failure, you get an actionable message:

Failed to start research — Google API error code 3 (DeepResearchErrorDetail).
This is likely a transient issue. Try again in a few minutes, or use --mode fast.

🔧 What Changed

• _extract_rpc_result() now detects all structured API errors in item[5], not just auth (code 16)
• start_research() refactored to use _call_rpc() — gets auth retry, server error retries, and debug logging for free
• New RPCError exception class with error code, detail type, and detail data attrs
• Service layer catches RPCError and provides user-friendly messages with actionable suggestions

📊 Tests

• 6 new tests (3 core, 3 service)
• 634 total tests passing, zero regressions
• Live end-to-end integration test: deep research start → status → import (54 sources) ✅

📦 Install / Upgrade

uv tool install notebooklm-mcp-cli
# or: pip install notebooklm-mcp-cli

Thanks to @isleofgreg for the detailed bug report! 🙌