I will next endeavour to complete the code for the ESP8266 module and then move on the the ESP32.
Inspired by the suggestion of running the AR488 sketch directly on the ESP32, when I looked at the board earlier in the year, I had only considered the number of available GPIO pins, which appeared to be sufficient for this purpose. The NodeMCU board could be powered directly from USB and hence 5v so it seemed a reasonable proposition.
However, it transpires from a further review of the ESP32 datasheet, that the ESP32 module, like the ESP8266, is a 3.3v device and hence works with 2.3v - 3.3v signals on its GPIO pins. The datasheet does not appear to specify the maximum current that can be handled by the GPIO pins, but under "Electrical Characteristics" in the "Absolute Maximum Ratings" table it does state that the "Maximum Drive Capability" is just 12mA. It would therefore seem that there may be insufficient current to drive the GPIB bus directly and even then, level shifting would be required on all 16 pins. If the level shifters could also be used to increase the available drive current then I suppose this might still be viable? The ESP32 should still be able to function as an add-on board connected via serial to any Arduino board that has spare Tx/Rx pins, but is it still a suitable candidate to drive the GPIB bus via its own GPIO pins? Does anyone have any thoughts on this?
In light of the above and depending on feedback, I might change priorities and look first at supporting additional layouts on the Mega 2560 including the double row end connector, and the ESP32/ESP8266 as add-on boards.