OPSIM-1093: feature. Migration from rubin_sim.

[yoachim]

Big migration of code from rubin_sim to rubin_scheduler including data, scheduler, site_models, skybrightness_pre, and utils along with relevant unit tests.

Several things from utils were deprecated (ObservationMetadata, time_handler, etc).

Dependencies on palpy were removed and coordinate conversions were swapped to use astropy.

The zernike tools in skybrightness_pre were not migrated as they had lots of palpy calls and unit tests that depended on outdated skybrightness files. It can be migrated later if wanted.

[rhiannonlynne]

There are a few files here still importing from rubin_sim, which needs either updates or moving or deleting.

scheduler/detailers/vary_exptime may be problematic (?)
scheduler/features/conditions.py just needs a docstring update
scheduler/survey/base_survey just needs a docstring update
scheduler/surveys/ddf_sched just needs a docstring update
scheduler/surveys/generate_ddf_grid may be problematic (??)
scheduler/thomson/thomson.jit.py - just an import update
scheduler/training/training.py - just an import update
scheduler/utils/sky_area.py just a few docstring updates
scheduler/utils/utils.py an update for rubin_scheduler.version
skybrightness_pre/data/generate_dark_sky.py just an import update
skybrightness_pre/data/generate_hdf5.py should probably just move to rubin_sim.skybrightness and formalize the expected output API for the hdf5 files / input API for skybrightness_pre

[rhiannonlynne]

It would be awesome if there was a way to more easily see what changed or not between the rubin_sim contents and the contents that moved here, but it seems like it's hanging together. A test example of the scheduler output between the two versions would be nice to confirm, not sure if you had other ideas.

[rhiannonlynne]

There are just a few more things to clean up, but this looks pretty good!

[rhiannonlynne]

Do we need to leave a test in rubin_sim?
I can’t tell if having this here will lead to problems with for example building docs, because rubin_sim won’t be available, but I guess we can try to solve that if it comes up. I agree that it’s nice to have it bundled here with the actual scheduler.