[abalakh]

Closes #1893.

Adds "Wake on lid open" to Battery Settings on the clamshells (RG SP, RG34XX SP, RG35XX SP): Off / 10m / 30m / Always. Off by default - keeping the hall sensor awake costs 5-6x the battery while asleep, according to the BaseOS dev, so it's opt-in.

10m/30m let the lid wake it for the first 10/30 minutes of sleep. After that an RTC alarm wakes it for a split second to switch back to Super Standby and keep the original shutdown timer, and from then on only POWER wakes it.

Tested on an RG SP (BaseOS 1.4.0): lid wake inside the window, no lid wake but POWER working after it, Off, and the shutdown timer surviving the switch.

[CatalyticArkun]

Follow-up update:
If the alarm that ends a 10m/30m window can't be set, lid wake is now switched off before returning. Before, os_sleep stayed at 0, with its higher battery drain, for the rest of the sleep.
xx_set_lid_wake keeps its policy value local (it was a leaked global _xx_policy).
PyUI's support check uses os.access(..., W_OK), matching the shell's [ -w ].
Tested on an RG35XX SP, BaseOS 1.4.0, with the card synced to this branch:

Off: the lid doesn't wake it; POWER does.
Always: the lid wakes it; POWER with the lid shut re-suspends.
10m: the lid wakes it inside the window. The window ends at +600 s, then only POWER wakes it.
The shutdown timer still fires at its original time after the window switch (15m timer: power-off at +15:02).
30m with a 15m timer and 10m with a 5m timer both take the single-suspend path, with lid wake on throughout.
13/13 checks passed, and the spruce log timings match. The failed-window-alarm path is stub-tested only (dash and bash).