Save spyglass version in analysis nwb

[samuelbray32]

          Unrelated to the immediate conversation above, but one thought from a conversation with @samuelbray32 is that we maybe should be recording the `spyglass.__version__` somewhere in the analysis NWB file so that we know when that was created. This would store the version, git hash and date which is nice. This wouldn't work retroactively, but certainly would help with forensics in a data analysis later if someone were trying to reproduce an analysis.

Originally posted by @edeno in #861 (comment)

[samuelbray32]

Was looking at this today. I think the best way to make sure this info goes in is to implement within AnalysisNwbfile.create() since it should be used universally when making an analysis nwb file. OPtions I see are:

1. Append the information onto the session_description key during AnalysisNwbfile.create() (easiest and most consistent across analysis types)
2. Create a new extension object in ndx_franklab_novela for the spyglass version info. Put this in during AnalysisNwbfile.create().
3. Add the spyglass version in the datatable description inserted during AnalysisNwbfile.add_nwb_object(). This may be stylistically better, but will have to search through the codebase for pipelines that don't use this function to insert their results (e.g. LFPV1() writes to the file using FirFilterParameters().filter_data_nwb() and already uses it's description entry for other information)

[edeno]

@rly any recommendations on where something like this might go in an NWB file?

[rly]

NWB has limited support for storing software versions right now. You can use NWBFile.source_script. It's a freeform text field. I would recommend storing either the version string or the github commit hash of spyglass there, the name of the function used, with either the name Spyglass or a URL to spyglass.

You could also create an extension as Sam suggested - that would be more structured than the above.