macOS support for multiple usage/usage-pages

[todbot]

Update of @fengji2006 PR#65, second try.

I think this includes all the requested changes, on top of the work done by @fengj2006 on #65. Unlike the previous attempt, I think this doesn't regress the other recent changes made.

You can now see all the different usages as different "devices" in hid_enumerate(). When running hidtest you see them too. On a Macbook Pro with internal keyboard/trackpad, there are many:

% ./hidtest | grep -E "(Product|Usage)"
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0x3 (0xff00)
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0xd (0xff00)
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0x2 (0x1)
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0x1 (0x1)
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0x5 (0xd)
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0xc (0xff00)
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0x6 (0x1)
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0x1 (0xc)
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0x6 (0xff00)
  Product:      Apple Internal Keyboard / Trackpad
  Usage (page): 0xb (0xff00)

Additionally, I've verified that an Arduino "RawHID" device with two usages shows up as two devices to hidapi. The device is a $9 Trinket M0 running TinyUSB pretending to be a Teensy in RawHID mode. See the "tinyusb_teensyrawhid_multiusage" sketch if you'd like to create a similar physical device to test against.

[Youw]

looks good overall

[Youw]

A small suggestion from my side:
When we do a squash merge, let's make a commit message clean.
E.g.: * apply requested changes and test is not informative in the master's commit history, so we better remove such entries.

[todbot]

Oh apologies. I debated what was the correct approach. Noted for next time. And thanks for being patient with me.

[Youw]

We're a team here :)