`DLCPosVideo`: Use provided epoch, multithread matplotlib

[CBroz1]

Description

This PR addresses #1160 by allowing the make_video process to continue without a PosIntervalMap entry.

This PR is also a partial address of #1157 by reverting to matplotlib as the default video processor, and saves the output as temp files. OpenCV as a tool for reading frames was very memory hungry (300gb for 10k frames) during testing of batch processing, and was probably memory leaking somewhere. Switching to a two step process that saves each frame to disk, RAM usage hovered ~7gb over baseline with the current batch size. More work could be done to dial in these defaults to optimize video creation time based on available resources

Checklist:

• [X] No. This PR should be accompanied by a release: (yes/no/unsure)
• [X] N/a. If release, I have updated the CITATION.cff
• [X] No. This PR makes edits to table definitions: (yes/no)
• [X] N/a. If table edits, I have included an alter snippet for release notes.
• [X] N/a. If this PR makes changes to position, I ran the relevant tests locally.
• [x] I have updated the CHANGELOG.md with PR number and description.
• [X] N/a. I have added/edited docs/notebooks to reflect the changes

[CBroz1]

Draft pending resolution of bug report from user