RFC: add filter function based enumeration

[Youw]

          We could do both: allow SDL filter out devices like this, _and_ avoid patching HIDAPI sources if we would introduce something like:

typedef int (*HID_API_CALL hidapi_ven_dev_filter_func)(unsigned short vendor_id, unsigned short product_id);
struct hid_device_info HID_API_EXPORT * HID_API_CALL hid_enumerate_ven_dev_filter(hidapi_ven_dev_filter_func);

But that is a different effort, and would definitely need an implementation on all platforms.

Originally posted by @Youw in #575 (comment)

[JoergAtGithub]

We filter devices in the postprocessing after enumeration. But it would make sense to define blacklist entries before enumeration. This makes als sense when we think about hot-plugging.
We currently use the following criteria:

• VID/PID to filter a few problematic devices
• Usage-Table/Usage to filter out whole device categories like mouses, joysticks and gamepads
• Bus-Type to filter out all embedded (SPI/I2C) laptop devices
• There is also one case where we use the interface number, but it seems this is only, because at that time Usage-Table/Usage detection wasn't reliable.

[Youw]

I agree all of those are valid points.

Knowing the codebase, I must say that to implement (early) filtering for each criteria will be challenging, e.g.: it is impossible to get the usage_pages/usages of the device with libusb backend w/o opening the device, which is so much for "early" filtering.

I believe on all (not sure about macOS - that one is a bit different) platforms, having an early filter for vid/pid only is simple-enough and probably should be our first step towards that direction.