[victormlourenco]

What

On the Mac, a controller that GameController reports as a Switch Pro Controller now rumbles through the controller's own rumble report 0x10, as SDL's hidapi driver does, instead of through Core Haptics.

Why

An 8BitDo pad in Switch mode (HID 057E / 2009, productCategory "Switch Pro Controller") never rumbled through Core Haptics on macOS 27.2. The log showed no error: the handle engines had no advanced players (NSCocoaErrorDomain 4097), so both motors were merged onto one Handles engine, and its plain players started and played. The pad still didn't vibrate. A small probe that wrote report 0x10 straight to the same pad made it vibrate, with or without first sending subcommand 0x48 (enable vibration). Why gamecontrollerd's output doesn't reach the pad is still open.

How

• pro_route (dm/host/haptics_apple.m) takes a free 057E / 2009 HID device that accepts a neutral report. The neutral report test skips the dead second entry IOKit can keep after a long Bluetooth disconnect (Mac: one Bluetooth Pro Controller comes back as two HID players after it reconnects, and Ocarina of Time reopens the controller applet in a loop (1,671 times), no input #2237). Log line: rumble through the Pro Controller rumble report (<transport>).
• The registry lookup the DualShock 4 route used is now hid_take, shared by both routes. The DualShock 4 route behaves the same; t_haptics_apple still covers it.
• Each of the game's two actuators keeps its own high and low band at the game's frequencies, encoded as HD rumble (pad_pro_rumble_side / pad_pro_rumble in dm/host/pad_pro.h, per dekuNukem's rumble_data_table.md). Rumble strength and Rumble cutoff apply as on the other routes.
• A rumble that is playing is sent again every 30 ms, as SDL does. A stop or a loss of focus sends the neutral side.
• Docs: a new entry, "Pro Controller rumble through its rumble report", in docs/reference/services.md; the symptom index regenerated; a line under "Found on this game" in the Ocarina of Time note.

Checked

• In game: The Legend of Zelda: Ocarina of Time (v131072, Mac app) rumbles with an 8BitDo pad in Switch mode over Bluetooth, over USB and through 8BitDo's 2.4 GHz receiver, on macOS 27.2.
• Host unit tests: make -C dm/host/tests GAME=0400486024DEC000 run passes. t_pad_pro gained encoder checks: the neutral side, the strongest full-amplitude side 00 C9 40 72, the frequency clamps, NaN input, the amplitude thresholds and their monotonicity, and the report layout. t_haptics_apple and t_haptics pass.
• Docs: dm/tools/tests/test_docs.py passes.
• C lint: dm/tools/lint_c.py compdb and tidy report no findings in the changed files. The 8 findings above the baseline are in files this PR doesn't touch (dm/gfx/tests/t_upload_src.m, t_vk_swapchain_life.m, t_wline.m, t_zero_depth.m, and dm/gfx/vk/hle_vk.m asking for a baseline shrink).
• regress.py: not run. The change only touches Mac controller rumble, which the boot, shader and frame checks don't exercise.

Not checked

• A genuine Nintendo Pro Controller, which rumbled through Core Haptics before and now takes this route too.
• Two Pro Controllers at once. GameController doesn't name a controller's HID device, so two that connect together may swap their rumble, as with the DualShock 4.