Only hold commanded values for VRF sub-units

[RobHofmann]

Problem

#508 (4.0.7) holds commanded values over the read-back for up to 8 s, so a VRF gateway's stale cache does not revert the UI. It does that for every unit, also for standalone units, which have no gateway cache.

That has a cost there, raised by @p-monteiro. When a standalone unit gets a value it cannot take, for example a swing mode it does not support, it corrects itself and reports a valid value. With the hold, the UI kept showing the refused value for up to 8 s. Before #508 it showed the corrected value at once.

Change

The hold and the 2 s follow-up read now only apply when the device is a VRF sub-unit, that is when the sub MAC differs from the gateway MAC. Standalone units behave as they did before #508: no hold, no extra poll per command.

The same change is in the 5.0 port, #522.

Testing

My fake-device harness drives the real climate.py. 39/39 checks passed:

• VRF sub-unit behind a stale gateway: unchanged. The UI keeps the commanded value, the 2 s refresh confirms it, a rejected command falls back after the TTL, and the refresh is cancelled on removal.
• Standalone unit: the command is confirmed on the first read, no refresh is scheduled, and a value the unit refuses shows the unit's own value on the next read.