Save fit params and develop rerun intercomparison tools

[Klankers]

Beta CTDCAL overwrites everything that gets generated in /logs and /pressure (and maybe more). For that reason, data analysts have historically needed to copy and paste the entire data folder, which leads to lack of disk space and data runs themselves require a lot of individual effort to compare results. To make this better on future analysts (as well as users tweaking their workflows), CTDCAL should support instanced runs and allow comparisons between runs without needing to move data around or rerun any processing steps.

• [ ] Add CLI option for "instanced", saving all data in outbox to timestamped folder (just change the save-to directory)
• [ ] Add CLI option and methods for comparing data within instanced folders. Particularly:
• [ ] Plotting results against each other, giving results on selected/all reported parameters
• [ ] Return differences in fitting coefficients (possibly formatting into a report with the formatted equation that was used)
• [ ] Return differences in flagging statistics, especially as functions of depth or bottle number

[DocOtak]

Check out https://github.com/xarray-contrib/datatree that is currently being integrated into xarray Slides from a presentation here: https://speakerdeck.com/tomnicholas/xarray-datatree-hierarchical-data-structures-for-multi-model-science