x86inc: Improve XMM-spilling functionality on 64-bit Windows

Allows for code like this:

    WIN64_SPILL_XMM 10, 14 ; allocates stack space for 14 xmm registers and spills xmm6-xmm9
    cmp            eax, ebx
    jg .foo
    ; code that uses 10 xmm registers
    RET
.foo:
    WIN64_PUSH_XMM  14     ; spills xmm10-xmm13 to the previously allocated stack space
    ; code that uses 14 xmm registers
    RET