Allow pausing xiaomi vacuum in all states

[pszafer]

I bought Xiaomii vacuum for my parents and found one bug and also reason for one annoying problem.
So two fixes in this PR.

For now user can only pause vacuum while cleaning. I thought that it was some limitation of python miio module, but no. It was component designed that way.

if self.state == STATE_CLEANING:
            await self._try_command(
                "Unable to set start/pause: %s", self._vacuum.pause)

so added STATE_RETURNING to this if.

2. When new vacuum is connected and no cleaning was performed xiaomi component is crashing.
   Don't know if checking should be added to every attribute update. Waiting for your opinions on it.
   For sure it was crashing on:

self.vacuum_state.clean_time
self.clean_history.count
self.clean_history.total_duration

Checklist:

• The code change is tested and works locally.
• Local tests pass with tox. Your PR cannot be merged unless tests pass
• There is no commented out code in this PR.

[syssi]

At least the second issue should be fixed at the underlying library (python-miio): rytilahti/python-miio#457

[pszafer]

ok, so reverted the second issue.

[pszafer]

@MartinHjelmare little out if this PR, but is this possible to get into this if ever? https://github.com/home-assistant/home-assistant/blob/6136ed31f24c921932c705e93be73b0e8935a36d/homeassistant/components/vacuum/xiaomi_miio.py#L301-L303

FAN_SPEEDS are not capitalized and we check capitalized string if it is in dict.

[MartinHjelmare]

FAN_SPEEDS are capitalized, with first character in capital letter.

[pszafer]

Haha, I lowercased it somehow in my head :)

[rytilahti]

This is intertwined with rytilahti/python-miio#471 -- what happens/what should happen when clicking start again after pausing?

[pszafer]

I don't think is that it is interwined.

About what should happen when clicking start again after pausing I imagine that default behaviour should be resume last state, not clean all.

[rytilahti]

One problem is that the state gets reverted (most likely) back to idle after some time (see xiaomi_miio.state implementation for an example of this happening with the error handling). Anyway, the backend library should definitely support resuming (unpausing) the state, this will be added as a new command.

The start_pause implementation as it currently is, is flawed. The description for that call is Start, pause or resume a cleaning task. so there needs to be more logic built into it to make it work as expected.

For the time being, could you check what mirobo raw-command get_clean_summary returns so I can shield the accesses to those variables properly?

[pszafer]

Output of command:

Sending cmd get_clean_summary with params []
[340436, 5692697500, 537, [1549201583, 1549190560, 1549107590, 1549100876, 1549053818, 1549052116, 1549051867, 1549018592, 1549011333, 1549010036, 1549009811, 1549008646, 1548928406, 1548920674, 1548919350, 1548919180, 1548917930, 1548917883, 1548917863, 1548917790]]

I think I can resolve conflicts in this PR and it can be merged to just repair first problem:

if self.state == STATE_CLEANING:

@rytilahti what do you think?

[rytilahti]

At that point the vacuum had cleaned already, so there's no reason for it to crash (like without one, which is likely fixed by my PR in python-miio). Anyway, this change can be merged after the conflict is fixed, there is AFAIK nothing wrong in calling the pause even when the device is already stopped.

[pszafer]

Rebased it. I think it is ready to be merged.