* use 'typedef uint8_t code_t' for char codes instead of uint16_t
* use 'typedef uint8_t height_t' for bitmap height instead of uint16_t, same for width, row, col
* optimize writing strings to and reading strings from SRAM (not used)
* delay initial display update after power on (not necessary but avoids two consecutive display updates when connecting the programmer (display update starts almost immediately after power on and restarts when programmer resets after ~1 second))