Prefer device_type over cluster/attribute values for Light

[puddly]

This is a test PR to potentially address home-assistant/core#164598 (comment).

I've added the affected devices to our test database and have regenerated diagnostics, showing what devices would be affected if we use the device type more aggressively. I haven't tested the changes yet but this affects very few devices.

[codecov]

Codecov Report

❌ Patch coverage is 66.66667% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 97.48%. Comparing base (c51efa5) to head (914b77d).

Files with missing lines	Patch %	Lines
zha/application/platforms/light/__init__.py	63.63%	8 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #703      +/-   ##
==========================================
- Coverage   97.51%   97.48%   -0.04%     
==========================================
  Files          62       62              
  Lines       10946    10954       +8     
==========================================
+ Hits        10674    10678       +4     
- Misses        272      276       +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[puddly]

Interestingly, there is not a single device in our testing database that uses a color device type (or a color temperature device type) without also having a color cluster! I would have thought this to not be the case but it looks like manufacturers use the device type pretty strictly compared to clusters themselves.

[puddly]

I've added the GLEDOPTO GL-SD-301P to diagnostics as well, from home-assistant/core#164598 (comment). It follows the same pattern: endpoint with a DIMMABLE_LIGHT device type and a bogus Color cluster with no capabilities. We erroneously identify it as a color temperature light. It somehow manages to send attribute reports or something to make ZHA change its color mode.

[TheJulianJES]

This is definitely useful, but I fear we'll break some lights. For a patch release, that's not great.

The main HA issue is that the light gets into a state where color_mode is set to XY (or HS?), even though that is not part of the supported color modes. As far as I can see, this can only ever happen during polling.

The initialization/capabilities set color_mode based on the supported ones, so that should be fine. For an alternative approach that just fixes polling for a patch release, see:

• #707

[TheJulianJES]

Checking Z2M and official instructions, the Sengled Z01-A19NAE26 does support color temperature. This patch now changes that to "brightness only". I imagine at least other Sengled lights will be affected as well.