Design issue in `Broker.async_update()`

[zxdavb]

@trvrnrth

I'd like you to consider this:

• Every update cycle, Broker.async_update() invokes self._update_device(device)
• instead, I believe that much of the code in that method should be in device.async_update()

So the code would become:

for device in self._devices + self._systems + self._zones + self._dhws:
    await device.async_device_update()  # instead of self._update_device(device)

Thus, in RamsesEntity.async_update() we can do I/O - e.g. we can address missing attrs, e.g. send_cmd(RQ|0004).

• although most of Broker._update_device(device) would end up in the base class RamsesEntity

Also has advantage that device.async_device_update() can be invoked for an individual entity without waiting for the next broker cycle.

[trvrnrth]

I'm not sure I follow entirely. In the code snippet, the device at that point is currently a ramses_rf Entity with _update_device just reading its state and updating a corresponding device in the HA device registry.

I wouldn't expect a ramses_rf Entity to become responsible for updating the HA device registry so are you suggesting adding a new ramses_cc Device wrapper around the ramses_rf Entity that would take on that responsibility?

[zxdavb]

Is this causing #170, #166?