Derived FSM style functions have lots of room for optimizations

[JulianKemmerer]

Want a pass that can combine sequential-looking FSM states that occur in the same cycle into one/as few combintatorial state transitions as possible (watching out for states that can't be jumped to / combined to others)

Then the parallel state grouping in C_AST_FSM_FUNDEF_BODY_TO_LOGIC and GET_GROUPED_STATE_TRANSITIONS seems like it has room for improvement, especially regarding ordering of inlined subroutine states vs single instance fsm module connection entry+FSM+exit triple state sequence.