fix(api): route requests to the correct regional Plaud host

[fcroiseaux]

Problem

Syncs stopped working with a Plaud sync failed: unexpected API response format notice. Plaud retired the global api.plaud.ai host in favor of region-specific hosts (api-euc1, api-use1, api-apse1), and each account's region is baked into its JWT. Calling the wrong host now returns an envelope with msg: "user region mismatch", which the client mapped to a fatal invalid_response — breaking every sync for anyone whose account is not on the (now dead) global host.

Root cause was confirmed from the live API response, not just docs: the failing call is listFiles() → GET /file/simple/web (the only unguarded call in runPlaudSync; per-file getFileDetail failures are caught and counted, so they can't produce the top-level "sync failed" notice). The captured envelope was { status: <non-zero>, msg: "user region mismatch" }.

Fix

• Regional host auto-detection. The API client probes the known regional hosts, sticks to the one that accepts the token, and reuses it for later calls. A new onHostResolved callback lets the plugin persist the resolved host so subsequent sessions skip re-probing (self-healing, and no manual region configuration required).
• Default domain moves from the dead api.plaud.ai to api-euc1.plaud.ai; the settings copy documents that the region is auto-detected (euc1 EU / use1 US / apse1 APAC).
• Listing query parameters (skip/limit/is_trash/sort_by/is_desc) are now sent to match the current /file/simple/web contract used by other Plaud clients.

Tests

• Added coverage for regional failover, resolved-host reuse (no re-probing), onHostResolved semantics, and the all-hosts-rejected error path.
• Updated the default-domain assertion.
• Full suite passes (59/59), npm run build typecheck is green.

🤖 Generated with Claude Code