Allow unicode up to 0xFFFFF using separate mapping table

[priyadi]

This allows unicode input up to 0xFFFFF using a mapping table in keymap file.

Some notes:

• To use, add UNICODEMAP_ENABLE = yes in Makefile
• Takes the range 0x7800-0x78FF in keycodes, hence a maximum of 2048 unicode characters are supported in a single keymap file.
• Need a mapping array in keymap file in the form of const uint32_t PROGMEM unicode_map[] = ...
• Uses the macro X() in keymap file, to refer to a value in mapping table.
• Could easily support up to 0xFFFFFFFF, but Linux only supports inputting 5 characters max, so I'm limiting to 5 bytes for now.

[algernon]

This appears to be doing something very similar to UCIS, perhaps it would make sense to use the same table UCIS does, so both functionality could benefit?

[priyadi]

@algernon I don't see how it could be possible. UCIS uses sequence of characters for input, this uses a single keystroke. Even if it is doable, it would require significantly more program space, and we still need to maintain a mapping table one way or another anyway. But, I'm open to suggestions, please see my keymap to see how this is implemented.

We should probably refactor the three unicode input methods, and separate the common functions to a file and make the different input methods to reside in separate files.

[algernon]

The way it enters the code is different, because I didn't get register_code32 working reliably. UCIS could, I think, use the same way of input you have written. And then, the only difference would be that with UCIS, the user types in the lookup key, and with your method, it's bound to a key instead. The same table could be used for both, but that needs UCIS to change how it enters chars...

Come to think of it, I could migrate UCIS to be based on your code, when this PR gets merged... then you don't have to worry about it at all :)

[jackhumbert]

Great! Thanks :)