Get AGB workflow running on HPC

[Aariq]

Maybe just needs more cores??

Things to try:

• Give main worker more cores. Maybe it needs additional memory to serialize objects??
• Allow workers to access data store directly rather than forcing main worker to handle all the passing of targets objects
• Run just esa_agb target as a test this works locally with 16gb RAM, just takes a long time
• Create run.R script that I can submit as a SLURM job so interactive session isn't needed to be up and running for many hours.
• Set wall time for heavy-duty workers much higher (like, 24 hours?)

[Aariq]

Bug in geotargets preventing me from using custom launchers for heavy-duty targets: njtierney/geotargets#66

[Aariq]

Might be good idea to split big datasets into tiles then do slopes on each tile with dynamic branching

[Aariq]

Haven't quite gotten it to work, but goign to close this and open a follow-up issue