add x86 instruction simplifier

[robertmuth]

The simplifier accepts an x86 instruction and returns a semantically equivalent but "better" x86 instruction. "better" usually means short encoding.

This would be implemented in CpuX64/ in a table driven fashion. The table itself can hopefully be derived more or less automatically from the existing x86 opcode table.

The feature would me used by CodeGenX64/codegen.py::_SimplifyCpuIns()

[robertmuth]

One way to implement this feature is to have a table that groups related instructions - roughly those with the same opcode prefix and then checks within those groups for better encodings.

The groups might also be helpful for translating from a more standard asm notation to Cwerg x86 instruction instructions.