why you are using port 0 (P0) for your output while it's multiplixed with address bus and you have port 1 (P1) free to use,i suggest you replace your code with P1.I hope this gives better data.

your wait_lcd subroutine should like that:

WAIT_LCD:

SETB EN ;Start LCD command
CLR RS ;It's a command
SETB RW ;It's a read command
MOV dataport,#0FFh ;Set all pins to FF initially
MOV A,dataport ;Read the return value
JB ACC.7,WAIT_LCD ;If bit 7 high, LCD still busy
CLR EN ;Finish the command
CLR RW ;Turn off RW for future commands
RET