Razer: mark the DeathAdder V3 Pro (0x00b6 wired, 0x00b7 receiver) verified on hardware

[EmreO33]

Summary

Hardware verification of the DeathAdder V3 Pro on both transports: stock HyperSpeed receiver 1532:00b7 and cable 1532:00b6 (firmware "Mouse 2.1", Windows 11), following the promotion procedure in docs/razer-testing.md. RazerHidClient was driven unmodified from Node over hidapi, with Razer Synapse and its services fully quit.

• src/razer/devices.ts: 0x00b7 and 0x00b6 get verified: true (comments record what was measured).
• src/drivers/razer/devices.test.ts: pinned in VERIFIED_SINCE and HARDWARE_VERIFIED.
• docs/razer-testing.md: id added to the verified list; the open "DeathAdder V3 Pro" section is closed out with the results, and the earlier 0x00b6 capture kept underneath as history.
• captures/razer-deathadder-v3-pro/: all feature reports of both write sessions (486 each) plus per-step results (serial payload redacted).

What was tested (receiver, 0x00b7)

Setting	Written	Read back on a fresh connection	After power cycle
DPI	800, 3200, 1600 (restore)	each value	1600
Polling (legacy command)	125, 500, 1000 Hz	each value	1000
Polling 2000 Hz	refused by the driver, as expected
Auto sleep	60 s, 900 s, 300 s (restore)	each value	60 s held
Low power	5 %, 50 %, 30 % (restore)	each value	5 % held

• Polling measured, not just read back: a pointerrawupdate counter (1 s window, peak) read 842 Hz at 1000 Hz and 124 Hz after the driver wrote 125 Hz. The legacy command really changes the link rate here, so highRatePolling: false is right.
• Sleep and low power persist: both re-read as written after switching the mouse off and on at its own power switch. That rules out the "no commit step / volatile state" theory the docs left open; the original report was made with Synapse present, which leaves contention as the explanation.
• Soak: 31/31 readStatus() reads over 3 minutes, 822 to 914 ms each, no stalls or errors.
• Read-only status: model name, "Wireless / HyperSpeed receiver", battery percent and state, DPI, polling, sleep, low power all decoded (steps 1 and 2 of the checklist).

Wired (0x00b6)

Same mouse with the cable plugged in, same scripts: 13/13 write/read-back steps passed and were restored; polling measured 126 Hz peak at 125 and well above 500 at 1000; battery reads "Charging" and climbed 25 % to 32 % across the wired runs; second 3-minute soak 31/31 at 824 to 887 ms. Settings written on one transport read back identically on the other.

Not covered

• Checklist step 8 (an off-list timeout set in Synapse appearing in the dropdown) needs Synapse running, which can't coexist with the test.
• Lift-off: the row keeps liftOff: false, so class 0x0b was not sent.

Tests

npm test: 1685/1685 pass before the wired commit; all Razer suites re-run after it, 150/150.