core: typedef'ed uint32_t to layer_state_t

[alex-ong]

This allows easy further modifications in future such as changing typedef based on number of layers.

[alex-ong]

Only part that may not be compatible when changing definition of layer_state_t is:

uint8_t shift = action.layer_bitop.part*4;
layer_state_t bits = ((layer_state_t)action.layer_bitop.bits)<<shift;
layer_state_t mask = (action.layer_bitop.xbit) ? ~(((layer_state_t)0xf)<<shift) : 0;
switch (action.layer_bitop.op) {
    case OP_BIT_AND: default_layer_and(bits | mask); break;
    case OP_BIT_OR:  default_layer_or(bits | mask);  break;
    case OP_BIT_XOR: default_layer_xor(bits | mask); break;
    case OP_BIT_SET: default_layer_and(mask); default_layer_or(bits); break;
}

since it relies on it being 32bits and references in 8 chunks of 4.

Possibly document what action.layer_bitop does, seeing as there are 6 parts to it?
bits(size=4) = the four bits.
xbit(size=1) = if enabled, mask = creates a mask of all 1's apart from the selected part
part(size=3, Values from 0->7) which bits to modify (first four bits, second set of four bits etc)
on(size=2) - not sure what this does yet
op(size=2) - operator - executes the given operator with (bits | mask)
kind(size=4) - action_kind_id

[tmk]

THanks.
I'll look into and test that part later.

Yea, that code shifts 4-bit nibble('bits') into plate that 'part' indicates.

    part            7    6    5    4    3    2    1    0                                           
    layer_state     0000 0000 0000 0000 0000 0000 0000 0000                                        
                    msb                                 lsb

shift(part*4) can be 28, 24, 20, 16, 12, 8, 4 or 0 when layer_sate is 32bit; 12, 8, 4 or 0 when 16bit and 4 or 0 when 8bit. We will have to check or truncate shift value depending on size of layer_state to prevent from over shifting like:

#if NUM_LAYERS > 16   // layer_state is unit32_t
    uint8_t shift = action.layer_bitop.part*4;
#elif NUM_LAYERS > 8   // unit16_t
    uint8_t shift = (action.layer_bitop.part & 3)*4;
#else  // uint8_t
    uint8_t shift = (action.layer_bitop.part & 1)*4;
#endif

[alex-ong]

Yup. I'll do changing NUM_LAYERS as a different request since this request stands fine on its own.
Since this pull request doesn't involve any functionality changes (since type-defs are all still uint32_t) it should be safe to merge.