Fix test_attributes_updated_not_replaced test

[TheJulianJES]

DRAFT. Testing.

Proposed change

• Update test_attributes_updated_not_replaced to validate inherited attributes using cluster._attributes_by_id instead of flat .attributes, so valid manufacturer-specific definitions sharing a ZCL ID are handled correctly.
  • Fixes false positive for: Let Hue sensor cluster inherit from OccupancySensing #4793
• Add a targeted guard that fails quirk-only attributes defined with manufacturer_code=None only when the ID collides with a base ZCL attribute, while allowing non-colliding IDs.
• Keep explicit ignore entries for known Sinope cluster issues.

Additional information

Device diagnostics

None.

Checklist

• The changes are tested and work correctly
• pre-commit checks pass / the code has been formatted using Black
• Tests have been added to verify that the new code works
• Device diagnostics data has been attached

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.65%. Comparing base (becb068) to head (42327c0).

Additional details and impacted files

@@           Coverage Diff           @@
##              dev    #4794   +/-   ##
=======================================
  Coverage   92.65%   92.65%           
=======================================
  Files         379      379           
  Lines       12745    12745           
=======================================
  Hits        11809    11809           
  Misses        936      936           

☔ 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.

[TheJulianJES]

The test didn't catch these before but they're really interesting:
Here, the default power_source attribute is overwritten (by name) with a manufacturer attribute and different attribute type:

zha-device-handlers/zhaquirks/sinope/switch.py

Lines 272 to 282 in becb068

	class SinopeTechnologiesBasicCluster(CustomCluster, Basic):
	"""SinopetechnologiesBasicCluster custom cluster ."""
	EnergySource: Final = EnergySource
	class AttributeDefs(Basic.AttributeDefs):
	"""Sinope Manufacturer Basic Cluster Attributes."""
	power_source: Final = foundation.ZCLAttributeDef(
	id=0x0007, type=EnergySource, access="r", is_manufacturer_specific=True
	)

And two ZCL attributes on the Metering cluster are overwritten here (by name as well) with a manufacturer attribute and different data type:

zha-device-handlers/zhaquirks/sinope/switch.py

Lines 294 to 302 in becb068

	class AttributeDefs(Metering.AttributeDefs):
	"""Sinope Manufacturer Metering Cluster Attributes."""
	status: Final = foundation.ZCLAttributeDef(
	id=0x0200, type=ValveStatus, access="r", is_manufacturer_specific=True
	)
	unit_of_measure: Final = foundation.ZCLAttributeDef(
	id=0x0300, type=UnitOfMeasure, access="r", is_manufacturer_specific=True
	)

There's a bit more discussion on that here: #4042 (comment)

Ultimately, this should not be allowed, right? Should it be possible for a quirk just to change the data type, e.g. do this without manufacturer_code or is_manufacturer_specific? Also no, right?

I guess we can just rename the colliding Sinope attributes and we're good?

[TheJulianJES]

There is another case we should consider:

zha-device-handlers/zhaquirks/bosch/rbsh_trv0_zb_eu.py

Lines 156 to 162 in a37b631

	pi_heating_demand: Final = ZCLAttributeDef(
	id=VALVE_POSITION_ATTR_ID,
	# Values range from 0-100
	type=t.uint8_t,
	zcl_type=DataTypeId.enum8,
	is_manufacturer_specific=True,
	)

Here, a ZCL attribute is overwritten with a manufacturer-specific one with the same name and a different data type, similar to the Sinope examples above, but the ID is also changed.
This should "mostly work", as ZHA reads attributes by name but it is likely still unexpected for a ZCL attribute to be replaced like this. We should look into validating this either directly in zigpy (but allowing exceptions?) or by improving this test.

For now, I'll try to see if I can add an exception for the Hue sensor in:

• Let Hue sensor cluster inherit from OccupancySensing #4793 (and unblock that PR)