Modify parsl multi_night_workflow to include filtering and results analysis

[wilsonbb]

The multi night workflow, which searches reflex-corrected WorkUnits across multiple nights should

a) filter out known objects using the ephemeris and a KnownObjMatcher
b) apply ML filter even for an empty model that keeps all results
c) saves an additional plots for analysis

We could make this part of the KBMOD search task since we already have the resampled WorkUnit and all of the results in memory.

We should also check that we're using an up-to-date ephemeris as part of the workflow. For ComCam, we just had a fixed cached ephemeris from SkyBot for the period.

[wilsonbb]

Some ideas for plots to add:

• the daily coadds
• a plot of the ImageCollection's footprints for each image overlapping a plot of our sky patch
• Plot of trajectories across mosaic as well as any known objects that were on the mosaic

[ColinOrionChandler]

As a NTS, I'm thinking of things that would require the Reprojected WorkUnit.

[wilsonbb]

#961 helps fix an error that was blocking this on the KBMOD side.

Next step will be updating and merging dirac-institute/kbmod-wf#62

[wilsonbb]

Doing an updated test of this, with new data