WIP: Finally Use Mffpy for read_raw_egi

[scott-huberty]

closes #6937
Maybe also #11380 ?
xref #8038

This is not ready for review(!) But I am using mffpy for a project and so I figured I would get this on the board..

Maybe a good target date for merge would be just after the MNE 1.9 release, so that devs can work with it for a few months before it is shipped in a stable release.

TODO:

• 1 Make it work
• Make it work with EEG data
• Make it work with EEG + PNS data
• Make it work with data with acquisition skips
• Re-integrate support for reading in channel status (e.g. bad channels)
• Get CI's green (specifically the MFF testing code).
• 2 Make it pretty
• Clean up the code
• Address all TODO/XXX comments
• Add tutorial for reading in calibration data directly with mffpy
• 3 RFC:
• Do we remove our custom reading code or keep it for legacy purposes?