Refactor of `exists_locally` (and consequently a lot of other stuff...)

[dafyddstephenson]

Branched from #74 , do not merge until after merging #74

The initial purpose of this PR was a short refactor of the exists_locally attribute to allow the case when a yaml file describing an input dataset existed locally, but the associated netCDF file did not. It kept unspooling from there with no clear finish line.

Ultimately, this is a lot of spring cleaning that is going to be necessary for the roms-tools integration. Changes to the user experience are fairly minimal (though have been updated in the example notebook), but what's happening behind the scenes is a lot cleaner and more intuitive, with hopefully all bodges and hacks removed.

closes #25

Summary:

• rename local_path attribute on InputDataset and AdditionalCode to working_path to avoid suggesting that the file(s) described by the path exist.
• Change InputDataset.get(local_dir) and AdditionalCode.get(local_dir) to actually fetch files to local_dir, not a convoluted subdirectory structure of it. Case.setup() now calls these methods with the required structure as an argument.
• Replace check_exists_locally on InputDataset and AdditionalCode with a property exists_locally which was not previously possible (and do the same for check_is_setup on Case, check_local_config_status on BaseModel)
• To make this possible, restructure of AdditionalCode in terms of expected directory/repo layout.
• ... Added an optional subdir attr/init argument to AdditionalCode in case the additional code lives in the subdirectory of a remote repository.
• Have changed the main example blueprint to remove MARBL additional code. Confusingly, as ROMS handles all input and output on MARBL's behalf, including reading things like namelist files, these are in fact ROMS additional code. There were lots of hacky things in the code to get around this simple fact just because it didn't fit in very nicely with the C-Star design concept. They are now gone.
• The main example blueprint now points to my fork of the example repo, which has been restructured, but I will update the C-Worthy version and C-Star blueprint once this PR is accepted. This was to avoid breaking tests for other people working with that repo.
• Provide explicit output_dir argument to Component.run() so it doesn't have to try and figure it out from the additional code directory structure anymore, which was ugly and confusing.

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[dafyddstephenson]

ok some stuff got chewed up in the merges but I think we're good to go now, have updated the diffs here