fix: raise ConfigEntryNotReady when amp is unreachable at startup

[steve28]

async_setup_entry let async_get_client's ValueError (device not detected, e.g. amp powered off) propagate as a raw exception, which HA treats as a hard setup failure with no auto-retry. Reload was required manually once the amp came back on.

Wrap the connection/probe logic and re-raise as ConfigEntryNotReady so HA's built-in retry-with-backoff picks the entry back up on its own once the device responds again.

fixes #23

[theharshl]

This response largely written by Claude but offer some decent detail.

Thanks for digging into this, @steve28, I appreciate you taking the time to submit a fix.

This issue is hopefully now fixed as of htd-home-assistant v0.0.36 (paired with python-htd v0.1.6), so I'm going to close this PR as superseded. The shipped fix takes a similar approach to yours, catching a connection failure in async_setup_entry and raising ConfigEntryNotReady so Home Assistant retries with exponential backoff, but scopes the catch more narrowly: the library now raises a dedicated HtdConnectionError only when the initial device probe/connect fails, and the integration catches only that specific exception rather than any Exception raised during setup. That keeps genuine bugs elsewhere in setup (e.g. in the source/zone name queries) visible as real errors instead of being silently swallowed into "retry forever."

One thing your PR got right that the shipped fix doesn't yet cover: wrapping the post-connect source/zone-name queries too, so a device that connects fine but drops mid-setup also retries gracefully. I've opened #25 to track that as a scoped follow-up.

Thanks again for the report and the fix! Closing this in favor of v0.0.36, but your PR helped confirm the right shape for the solution.