Revert "Delay appending the HID descriptors until .begin()"

[algernon]

Reverts #62, because it causes issues on the Dygma Raise. With this applied, the keyboard either doesn't register:

[  +0.110901] usb 2-3.4.2: New USB device found, idVendor=1209, idProduct=2201, bcdDevice= 1.00                                                    
[  +0.000006] usb 2-3.4.2: New USB device strings: Mfr=1, Product=2, SerialNumber=3                                                                
[  +0.000003] usb 2-3.4.2: Product: Dygma                                                                                                          
[  +0.000003] usb 2-3.4.2: Manufacturer: Raise                                                                                                     
[  +0.000002] usb 2-3.4.2: SerialNumber: BF00F8E650515157382E314AFF062B0BCkbio01                                                                   
[  +0.012923] cdc_acm 2-3.4.2:1.0: ttyACM0: USB ACM device                                                                                         
[  +0.004397] input: Raise Dygma as /devices/pci0000:00/0000:00:14.0/usb2/2-3/2-3.4/2-3.4.2/2-3.4.2:1.2/0003:1209:2201.0010/input/input37          
[  +0.064089] hid-generic 0003:1209:2201.0010: input,hidraw4: USB HID v1.01 Mouse [Raise Dygma] on usb-0000:00:14.0-3.4.2/input2                   
[  +0.005623] usbhid 2-3.4.2:1.3: can't add hid device: -75                                                                                        
[  +0.000020] usbhid: probe of 2-3.4.2:1.3 failed with error -75

Or, when using the driver/hid branch of Kaleidoscope (keyboardio/Kaleidoscope#771), the keyboard registers, but whenever I press a key:

[  +1.738457] usb 2-3.4.2: input irq status -75 received

If that -75 is -errno, then it's EOVERFLOW: "Value too large for defined data type".