Feat/kathleen improvements

[eKathleenCarter]

made small changes to the llm_submit.py function:

1. Chunking now takes in to consideration the max token size of the model selected
2. Output is saved at the end of each succesful chunk (in the case of running out of credits or failing mid process)
3. If failure happens, the function can now pick up from where it left off.
4. all results are combined at the end of a successful run and the per-chunk output is cleaned up.

[waTeim]

Cool, but can you merge it to develop instead of to main?