ASUS: ROG OMNI receiver mice and keyboards (Harpe Ace Mini, Falchion RX Low Profile), unit tests

[metalgitter]

Adds the ROG OMNI receiver (0B05:1ACE) to the ASUS provider: the battery of the mouse and the keyboard paired to it. It was listed as "not included" in providers/asus.py.

Closes: no issue (none open for the OMNI receiver).

Source of the protocol: G-Helper (seerge/g-helper), app/Peripherals/PeripheralsProvider.cs (DedectOmniMouse, MouseFromOmniPid, KeyboardFromOmniPid) and the *Omni model classes in Mouse/Models and Keyboard/Models (AsusKeyboard.ReadBattery, Falchion.ParseBattery). Only the protocol facts are used.

How it is read: the receiver's interface 0 is only a keyboard, so the existing interface-0 path never matched it. The vendor collections are on interface 2, each with its own report id and 64-byte reports:

Collection	Request	Reply
col01 ff02	01 a0 00 00	pair list: from byte 5, four bytes a slot, pid little-endian, 0 ends
col03 ff01	03 12 07	the usual 12 07 mouse reply behind report id 3 (battery byte 5, charging byte 10)
col02 ff00	02 12 01	keyboard: battery byte 6 (byte 11 on the Falchion family), charging byte 9

Only pids in the pair list that G-Helper knows are asked. Collections are grouped per receiver by device instance, so two receivers do not mix. An OMNI mouse uses the same names as KNOWN, so it shares one icon with the same mouse on its cable. Keyboards get kind="keyboard". A sleeping device (all zeros, battery 0 without charging, or ff aa from a keyboard asleep on battery) shows nothing. _read now uses a small _ask helper shared with the OMNI requests; the existing mouse behaviour and its tests are unchanged.

Tested on real hardware: yes, a ROG Harpe Ace Mini (pid 1b65) and a ROG Falchion RX Low Profile (pid 1b06) on one OMNI receiver. --probe shows:

[ASUS] ROG OMNI receiver pid=1ace
    reply: 01 a0 00 02 00 06 1b 02 04 65 1b 03
  paired: 1b06 1b65
  mouse ROG Harpe Ace Mini pid=1b65
    reply: 03 12 07 00 00 37 02 0a e8 0e 00 00
  keyboard ROG Falchion RX Low Profile pid=1b06
    reply: 02 12 01 00 00 00 42 02 01 00 14 42
  -> ROG Harpe Ace Mini: 55%
  -> ROG Falchion RX Low Profile: 66%

The keyboard answered 02 ff aa while asleep and 66 % once awake. The other OMNI pids (Keris II Ace / Origin, Harpe Ace Aim Lab / Extreme, Gladius III Aimpoint, Keris Wireless Aimpoint, Strix Impact III Wireless; Azoth, Azoth Extreme, Strix Scope II 96) come from G-Helper and are unverified.

Unit tests: 16 new tests in tests/test_asus.py (pair list, mouse and keyboard parsing, the Falchion gauge byte, standby, events before the reply, unknown pids, two receivers, OMNI + cable sharing an icon). python -m unittest discover -s tests: 475 tests OK.

Other open pull requests changing the same files: none touch providers/asus.py or tests/test_asus.py. Most open PRs also add to CHANGELOG.md, README.md and docs/protocols.md, and #181, #177, #176, #172, #167, #157, #141, #137, #136, #135, #134, #129, #124, #123 and #39 change halo_battery.pyw; this PR only changes the "asus" Device types label there and a line of the module docstring.

🤖 Generated with Claude Code