Raise a proper error when the bridge can't be reached

[michaelarnauts]

Opening the connection only translated asyncio.TimeoutError:

try:
    self._reader, self._writer = await asyncio.wait_for(asyncio.open_connection(self.host, self.PORT), TIMEOUT)
except asyncio.TimeoutError as exc:
    raise AioComfoConnectTimeout("Timeout while connecting to bridge") from exc

Everything else that asyncio.open_connection() can raise came out as a bare OSError: a refused connection, a bridge that is gone from the network, a hostname that no longer resolves. That is not something a caller can reasonably catch, and it is what the reporter of home-assistant-comfoconnect#100 ran into:

OSError: [Errno 113] Connect call failed (192.168.178.142, 56747)

It also ended up in the catch-all handler of the reconnect loop, which logs the full traceback, every five seconds for as long as the bridge stays away.

OSError is now translated to a new AioComfoConnectNotReachable, and the reconnect loop retries it the same way it retries a timeout, with a one line warning instead of a traceback.

Some notes:

• The TimeoutError handler has to stay above the OSError one, since TimeoutError is a subclass of OSError. There is a comment saying so.
• ComfoConnectNotReachable already exists but means something else, the gateway telling us it cannot route a request to a node, so this needs its own exception. It is in the Aio family with the other connection level errors.
• The new tests cover an unreachable host, a refused connection and a name that does not resolve.

For the Home Assistant integration this needs a small follow up: it catches AioComfoConnectTimeout to trigger the rediscovery that handles a bridge that changed address, and AioComfoConnectNotReachable should trigger the same path.

🤖 Generated with Claude Code