Add InputObservations

[moeyensj]

This PR introduces a new table InputObservations that represents de-normalized combinations of detections and their exposure information (such as those available in AIMS and commonly available in other places):

import quivr as qv
from adam_core.time import Timestamp

class InputObservations(qv.Table):
    id = qv.StringColumn()
    exposure_id = qv.StringColumn()
    time = Timestamp.as_column()
    ra = qv.Float64Column()
    dec = qv.Float64Column()
    ra_sigma = qv.Float64Column(nullable=True)
    dec_sigma = qv.Float64Column(nullable=True)
    ra_dec_cov = qv.Float64Column(nullable=True)
    mag = qv.Float64Column()
    mag_sigma = qv.Float64Column(nullable=True)
    filter = qv.StringColumn()
    observatory_code = qv.StringColumn()

The Observations table has a new constructor Observations.from_input_observations(input_observations) that maps the input observations into the format used by THOR, namely by link_test_orbit. The THOR observation table has been modified to use SphericalCoordinates instead of PointSourceDetections.

The transition to spherical coordinates is convenient since operations are more often than not done on coordinate classes and not PointSourceDetections.

This PR also adds a Photometry table used by THOR observations which I anticipate will one day be added to adam_core for use in things like light curve fitting.

[moeyensj]

Tests and linting pass locally so I'll merge this. Looks like there is an upstream issue getting the OBSCODE file from the MPC.