fix(calendar): propagate todo merge failures

[jiahao6635]

Fixes #1252.

Summary

• Remove the catch-all VTODO merge fallback that rebuilt a todo from the partial update payload.
• Let unexpected merge failures propagate through update_todo before todo.data is replaced or saved.
• Add an async regression test that injects a non-ValueError failure and verifies the stored data is unchanged, save is not called, and no rebuild occurs.

Why

Rebuilding from a partial update silently drops every stored property the caller did not resend while still reporting success. Surfacing the original failure preserves the existing VTODO and matches the event merge behavior.

Testing

• The new synthetic RuntimeError test fails on base 9621ffa because the exception is swallowed.
• uv run --frozen pytest tests/unit/client/test_calendar.py -q: 70 passed
• uv run --frozen ruff format --check nextcloud_mcp_server/client/calendar.py tests/unit/client/test_calendar.py
• uv run --frozen ruff check nextcloud_mcp_server/client/calendar.py tests/unit/client/test_calendar.py
• uv run --frozen ty check -- nextcloud_mcp_server

[CLAassistant]

All committers have signed the CLA.

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud