Refactor ministack planning, create MiniStackPlanner and refactor internals of

[scottstanie]

Motivations were:

1. Start to separate the functionality of reading the input stack from planning out which ministacks will be created (it had been done inline in sequential.py and/or in the VRTStack, which should just read).
2. Fix the problem that, when running one ministack with a compressed SLC at the start, we don't make the right interferograms
   • it would always use only real SLC dates in the interferogram step
3. Fix the problem that we had no way of knowing what the "base phase" was for for single ministack runs if we didn't start from the beginning with sequential.py

Since these were all interrelated, it ended up being a large refactor. Also since so much of the date/file logic was being touched, I fixed the other lingering problem of using date everywhere internally when there was no reason not to use datetime (which would be necessary for any intra-day repeat mission).
(1) and (3) are done, and (2) is set up so it'll be a smaller upcoming PR.