hi,
well, if you use just two bits, so:

- eject memory allocations for BIT2_DELAY...BIT7_DELAY;

- in main/next bit loop use
CJNE R7,#2,NEXT_BIT ; ...to 1
instead of
CJNE R7,#8,NEXT_BIT ; ...to 7
or just optimize it to simple:

  MOV R7,#0 ; process bit 0
  CALL PROCESS_BIT
  INC R7    ; process bit 1
  CALL PROCESS_BIT

- do not produce BIT2_DESCRIPTOR...BIT7_DESCRIPTOR
and so do not include their addresses into lookup table.

That`s all.

Regards,
Oleg