di-cal and subtract with external model.

[d3v-null]

This would allow model visibilities to be generated in Woden.

Would also allow chunks to be done in multiple vis-simulate calls, and combined separately, as an alternative to #9

[d3v-null]

Bonus points for calibrating from model column of measurement set to allow for wsclean self-cal

[JLBLine]

This may be more confusing than helpful, but this code here https://github.com/ADACS-Australia/PyFHD/blob/main/PyFHD/source_modeling/vis_model_transfer.py is how I setup PyFHD to read WODEN models and match to real data. The annoying thing being WODEN doesn't flag tiles, so if you're calibrating a data set which has missing tiles, you have to do a bunch of book keeping to work out which visibilities in the model match what's in the data.