PiCamera class and lil inter-version dev update

[sneakers-the-rat]

Hello! I've drafted a controller to PiCamera/Raspi Camera Modules (finally), it should be pretty fast as it offloads all the frame processing to the gpu (using the wonderful picapture) and reads the frames directly to a numpy array from the preallocated frame buffer. You can see it here (and in the parallax branch) https://docs.auto-pi-lot.com/en/parallax/autopilot.hardware.cameras.html#autopilot.hardware.cameras.PiCamera Haven't had a chance to test it yet because working on many things in parallel, but should work!

I've been pretty scattered as I'm working on getting rolling on the last experiment for my dissertation as well as writing another... uh... manifesto, and so I haven't been able to cohere changes across branches. Here's the general status of things that will be rolled up in v0.4.0

• dev - the general development branch is what all changes will be pulled to before bringing back to main. That has some gui refinements from @cxrodgers that need to be merged
• parallax - the main 'hot' development branch i'm building an experiment on right now, has a bunch of refinement on a few hardware devices and some new geometry objects.
• registries - outstanding work (that's almost done!) on implementing @arnefmeyer 's excellent registry system for better separation of user and library code, that issue is here Implement Registries - New protocols not visible #39
• subject_cleanup - the absolutely cursed Subject class needs a makeover. I got that started, and will be working on making moves towards implementing neurodata without borders as a backend at the nwb hackathon next week.

The rest of the issues flagged for v0.4.0 are in this milestone: https://github.com/wehr-lab/autopilot/milestone/1

Every bit i work on, smoothing out the first rough draft of the library, makes me love working with it even more :) still lots of bugs and rough stuff, but one day we'll have the magic tool of our dreams!