feat(ch_ephem): CHIME ephemeris refactored from ch_util.

[ketiltrout]

Extracts a lot of ch_util.ephemeris into a separate package.

General changes:

• Solar and lunar ephemeris functions have been converted into caput.time.Observer methods: feat(time): Solar/lunar ephemerides radiocosmology/caput#271
• Other functions that take a Observer object, now use None as a default, and load the chime observer during execution if necessary. (This is to avoid unnecessarily importing ch_ephem.observers.)
• Could use a test-suite, but this PR is big enough already.

More comments on specific submodules follow.

Catalogs:

This contains all the ch_util/catalogs plus a function (ch_ephem.catalogs.load) to read them and return the parsed JSON.

The standard source catalogues loaded by the sources submodule need to move to ch_ephem. The other catalogues could stay in ch_util, but it seems simpler to me to keep them all in the same place and provide a uniform interface to them.

Coord:

Co-ordinate transforms etc. These are essentially unchanged from their versions in ch_util.ephemeris.

Observers:

Re-creates all the caput.time.Observers which used to be created by ch_util.ephemeris, but for the most part, they have to be fetched via name from the get() function. The exception to this is the chime Observer, which can be directly imported, since it's the one usually needed.

Position data for the observers is now in a YAML data file instruments.yaml which is parsed at import time. New instrument Observer objects can be added to ch_ephem just by adding information for it to the YAML file.

Also note that Observers returned by this submodule have an extra attribute, rotation, inserted into their __dict__, which is not in caput. This rotation is needed by ch_ephem.coord.peak_ra. If we don't want to do this to these Observers, one other options would be to keep peak_ra in ch_util (moving it to, say, ch_util.tools), and then there's no need for rotation information in this module.

Like sources, this submodule reads data from disk at import time, so is not automatically imported into the base ch_ephem object.

Pointing:

Just Galt telescope pointing routines, unchanged from ch_util.ephemeris. These are another candidate we could consider putting somewhere else in ch_util, if we don't want them here.

Sources:

Loads the standard source catalogs, and provides the calibrator objects CasA, CygA, TauA, VirA. Like observers, this submodule reads data from disk at import time, so is not automatically imported into the base ch_ephem object.

Time:

Time utilities, largely unchanged from what was in ch_util.ephemeris.

[ketiltrout]

To pull out one major question I had from the description that I'd like opinions on:

• Should the rotation data be in ch_ephem? Or would it be better to keep it in ch_util and move the peak_ra function there? Or just delete peak_ra and keep rotation in ch_util?

[ljgray]

To pull out one major question I had from the description that I'd like opinions on:

• Should the rotation data be in ch_ephem? Or would it be better to keep it in ch_util and move the peak_ra function there? Or just delete peak_ra and keep rotation in ch_util?

I don't mind moving the rotation data here, personally.

[ketiltrout]

I've added ruff to the linting (which found all my typos) and move the rest of the geometry from ch_util.tools (roll and offset) over here, too.