Python SDK does not robustly handle BLE connection failures / drops

[epheo]

Note! This ticket will capture any and all issues relating to Python SDK BLE:

• recovery from failures to establish connection
• recovery from connection drops
• closing and re-opening Wired / Wireless GoPro objecvts

In general, these use cases have not yet been designed for / tested against and I'm not sure much of this is even solvable in the Python SDK due to the lack of robustness in Bleak and the OS's BLE stack. But it needs a major investigation on each operating system.

Component
In the Python SDK.

Describe the bug
If the connection to a GoPro fails, the WirelessGoPro._periodic_keep_alive coroutine is left open.

To Reproduce
Start connection to a GoPro and have it fail (stop ble adapter or remove GoPro battery)
Then we can observe _periodic_keep_alive by checking the running coroutines:

tasks = asyncio.all_tasks()
coroutine_name = task.get_coro().qualname

Expected behavior
I would expect the keepalive to be stopped once the Connection fails.

Hardware
Hero 11
Linux 6

Additional context
I'll try to investigate later if this can be solved by initiating connection within a contextlib.AsyncExitStack().

[tcamise-gpsw]

I've made some changes to this in the last release. However the bigger issue here is that the Wire(less)GoPro objects are not designed to persist through subsequent connections. I haven't spent much time on this because I don't think bleak robustly manages, at least on Windows, failed / reconnections.

In any case, there is a large investigation required to improve the robustness here.

[tcamise-gpsw]

I'm going to use this ticket as the gathering location for all BLE connection-robustness issues.