Search features requested

[nocollier]

• Time bounds in search: While search is used to identify relevant datasets which have no specific time extent beyond the definitions of the corresponding experiment, users want to be able to include a start/end time to only download files on a span that is relevant to them. For coarse output (monthly/annual), modeling centers tend to group everything in a single file, but for daily data they are frequently broken up. Allowing specification of time bounds at search would keep them from downloading huge files they will not use. Does either index type support this?
• Add a form of negation: Sometimes users want to find, say, all the models except a certain few. Both Solr/Globus APIs support this, we just need to settle on a python interface. source_id=['!CESM2-WACCM']? not_source_id=['CESM2-WACCM']?
• Autofill search: Could we give an API so users could type facet values like historical tas pr mon r1i1p1f1 and we could lookup matching facets and autofill? experiment_id='historical', variable_id=['tas','pr'], frequency='mon', variant_label='r1i1p1f1'

[nocollier]

Regarding time bounds. According to Sasha, the metadata records are not consistent across all file records. We could provide keywords that handle this in the search, but not use them in the index node queries, and then apply them when downloading files by parsing the time out of the filename.

Max and I settled on providing these options: time_start, time_end, time=slice()

We can add negation to where these symbols will be accepted: !, !=, ~ for negation