Retries behaviour for cloud based (or not) integrations

I’m having constant issues with integrations that need to talk to ip-based devices (local network devices, internet vendor APIs, etc):
The integrations appear as “could not configure” or the devices as unavailable, and NEVER go back to available.

The reason could be that Hass rebooted as my router was down, so it could not connect to the remote API or to the local device. It could also be that the devices were temporarily unavailable.

For the integrations to be available again, I need to open the integration’s small menu and click on “Reload”.

But IMHO when devices / integrations can’t connect, they should report their state as “should retry later”, and Hass should “poll” the remote api/device and automagically reload the integration when possible !

NB: The issue does not appear when the API/device is unavailable AFTER the integration was loaded. The issue really is about when the initial integration loading fails.