Ticket/2453/bad/mouse

[danielsf]

This PR addresses

#2453 by making our _filter_on_death_date code treat death_on = None as "the mouse is still alive" (though that mouse was dead and we have updated LIMS accordingly)

#2452 by adding a failed_ecephys_session_id_list field to the metadata writer schema. These sessions will be used when constructing each mouse's history with our apparatus, then dropped at the moment of output.

[danielsf]

By way of validating that this actually addresses #2452, I re-generated the metadata tables using this branch. The passed sessions for the mice the Corbett called out are all marked as session_number = 2 now

>>> import pandas as pd
df>>> df=pd.read_csv('ecephys_sessions.csv')
>>> df.loc[df.mouse_id==524760, 'session_number']
1    2
Name: session_number, dtype: int64
>>> df.loc[df.mouse_id==546512, 'session_number']
48    2
Name: session_number, dtype: int64
>>> df.loc[df.mouse_id==5484865, 'session_number']
Series([], Name: session_number, dtype: int64)
>>> df.loc[df.mouse_id==548465, 'session_number']
54    2
Name: session_number, dtype: int64

[morriscb]

looks good to me.