Modify StimulusFile to accept inhomogenious formating from VBO

[morriscb]

A change was made to the SDK [remind me the ticket @danielsf] that pulled data directly from the StimulusFile instead of from LIMS as the data on LIMS was found to be unreliable. This breaks some behaviors in VBO as the code as implemented assumes a standard data structure and the VBO data does not follow a fixed format. This issue shows up when trying to use get_ophys_experiment_table from the LIMS API.

The steps to address this on this ticket will be as follows:

• [ ] Wrap a try/except with a pdb around the validation method for the stimulus file to catch sessions with failing stimulus files one at a time.
• [ ] Modify the validate code and the StimulusFile code to handle these non-standard stimulus pkl files and be usable within the standard object.
• [ ] Repeat until all sessions for VBO pass and sessions from VBN remain intact/workable.

[danielsf]

The PR that makes "read session_type from the pickle file" default is #2488

The idea of reading this data from the pickle file probably first comes from #2426

[morriscb]

We're currently waiting on tickets from the LIMS team that are correcting the information that necessitated switching to the pkl files. Assuming this gets merged, we may not have to worry about this issue. However, depending on when they are merged we may have to work on this ticket.

[morriscb]

Expect that session type and date of acquisition are the main issues here. If we can develop a method to infer both from data within the stimulus file we can 1) add that to our code as a test to handle these cases or 2) provide the example to the platform team so they in turn can add the session type and correct daq to LIMS.

[morriscb]

For the data to be released with VBO this is not an issue. Only for other significantly older data.