Rethink result caching in library

[SRGDamia1]

The library has code for dumping and re-reading json files with results. I was using that to make it easier for myself to combine batches of analysis and modeling runs when I just wanted to add a few or re-run a few but I wanted everything to end up in the same end dataset. Basically - I was creating fall-back's for myself so if I submitted 100 sites to run and 87 ran ok and then it crashed I could just restart the same job and it would read those first 87 results from saved jsons and submit requests to ModelMW when there wasn't a saved file. This is probably sloppy or bad form.