Cloud source: more resilient to slow or failing endpoints
Fixes
- Every Tigo cloud request is now capped at 30 s. Previously the Home Assistant shared session default applied (300 s), so a single slow endpoint could stall a whole poll for minutes (#41).
- A transient failure (timeout, network error, HTTP 5xx) of a single endpoint no longer fails the whole update: that endpoint keeps its last values while the others refresh normally, so sensors no longer go
unavailablebecause of one slow endpoint (#41). - The update only fails, and backs off, when every endpoint called in the cycle fails. Authentication errors, throttling (429 / Retry-After) and 4xx responses are handled as before.
- A failed energy fetch is retried on the next cycle instead of waiting for the next energy interval.
- Network timeouts are now logged with a readable reason (
TimeoutError()) instead of an empty message.