Ndx fiber photometry

[pauladkisson]

Fixes #21

See example session

[alessandratrapani]

General comments:

1. Be consistent in naming all the variables and fields that refer to the isosbestic path. You call it either isosbestic or reference. I suggest going with isosbestic (I add a few Suggested Changes but I'll leave you the full revision).
2. It was suggested to me on old reviews of my PRs to name all the objects in the NWBFile keeping the nomenclature of the class they refer to, e.g instead of "dms_signal_excitation_source" --> "ExcitationSourceSignal". I tried my best to suggest the changes for all the devices and others, but I missed something for sure. I'll let you correct the rest.
3. The scope of this new extension was to share resources as much as possible. Try to avoid duplicating devices that share manufacturer and model.

[pauladkisson]

It was suggested to me on old reviews of my PRs to name all the objects in the NWBFile keeping the nomenclature of the class they refer to, e.g instead of "dms_signal_excitation_source" --> "ExcitationSourceSignal". I tried my best to suggest the changes for all the devices and others, but I missed something for sure. I'll let you correct the rest.

There is some open discussion on this: rly/ndx-pose#31
And given all the considerations mentioned on that issue, I think snake_case is the more appropriate/intuitive choice for naming, so I'm going to stick with it for this conversion.

[pauladkisson]

The scope of this new extension was to share resources as much as possible. Try to avoid duplicating devices that share manufacturer and model.

I see. I was using data objects to represent the physical devices (ex. 1 ExcitationSource object per different LED source). But it sounds like the idea is to collapse the metadata as much as possible, which is fine by me.

[alessandratrapani]

LGTM