Provide a planar screen as an optional assumption for handling off-disk helioprojective coordinates

[ayshih]

Expanding beyond assume_spherical_screen() (see #4003), this PR adds a planar screen as an optional assumption for handling off-disk helioprojective coordinates.

See also #6411

[wtbarnes]

OK I've taken a crack at implementing the screens as separate classes and modifying the context manager and make_3d such that they can handle any of these screen classes and all of the logic of those screen assumptions is in those particular classes.

Some things to still left to do/decide

• Is this the API we want? Should we instead make the classes themselves the context managers? If the latter, I'm not sure how to handle modifying the class attribute _assumed_screen in the case where the context manager is not a class method
• Implement the context manager on the classes themselves
• Make BaseScreen an abstract base class
• assume_spherical_screen should be deprecated in favor of assume_screen
• Docstrings need to be updated
• Examples need to be updated
• Modify PlanarScreen to allow for a variable distance from the center of the Sun
• Changelog
• Understand what happens when you're looking away from the planar screen
• Replace assume_spherical_screen with SphericalScreen everywhere

[wtbarnes]

How this now works:

import astropy.units as u
from astropy.coordinates import SkyCoord

from sunpy.coordinates import Helioprojective, SphericalScreen, PlanarScreen

frame = Helioprojective(observer='earth', obstime='2020-01-01')
coord = SkyCoord(Tx=[-1000, 300, 1000]*u.arcsec, Ty=[-1000, 300, 1000]*u.arcsec, frame=frame)

with SphericalScreen(coord.observer):
    _ = coord.make_3d()

with PlanarScreen(coord.observer):
    _ = coord.make_3d()

with PlanarScreen(coord.observer, distance_from_center=1*u.Rsun):
    _ = coord.make_3d()