[BUG] Allow smoothing of derived SPH fields onto covering grids

[jzuhone]

PR Summary

The following script fails on main:

from yt import load
bbox = [[40669.34, 56669.34], [45984.04, 61984.04], [54114.9, 70114.9]]
ds = load("TNGHalo/halo_59.hdf5", bounding_box=bbox)
w = ds.quan(0.2, "Mpc")
le = ds.domain_center-0.5*w
re = ds.domain_center+0.5*w
g = ds.r[le[0]:re[0]:128j,le[1]:re[1]:128j,le[2]:re[2]:128j]
print(g["gas","temperature"].min())
print(g["gas","kT"].min())

with this error:

  File "/Users/jzuhone/fix_grid.py", line 9, in <module>
    print(g["gas","kT"].min())
          ~^^^^^^^^^^^^
  File "/Users/jzuhone/Source/yt/yt/data_objects/data_containers.py", line 235, in __getitem__
    self.get_data(f)
  File "/Users/jzuhone/Source/yt/yt/data_objects/construction_data_containers.py", line 899, in get_data
    self._fill_sph_particles(fields)
  File "/Users/jzuhone/Source/yt/yt/data_objects/construction_data_containers.py", line 977, in _fill_sph_particles
    raise KeyError(f"{ptype} is not a SPH particle type!")
KeyError: 'gas is not a SPH particle type!'

It appears that derived fields were not supported for this operation. This PR addresses this issue by making extra checks to determine if a field is either an SPH on-disk field, derived field, or a particle filter type of an SPH field.

However, in order to avoid many more onerous code changes, I had to provide for the creation of fields such as ("gas", "particle_position"), ("gas", "particle_mass"), and ("gas","particle_position_[xyz]"), which currently do not exist because the SPH position fields are aliased to ("gas", "[xyz]") and the mass field is aliased to ("gas", "mass"). Happy to discuss if this doesn't feel right to anyone.

PR Checklist

• New features are documented, with docstrings and narrative docs
• Adds a test for any bugs fixed. Adds tests for new features.

[matthewturk]

So I am OK with this, and I think it's OK to go in as-is, but I did have two requests that might help future maintainers.

[matthewturk]

Can we change fi to finfo for this function?

[jzuhone]

Done

[matthewturk]

I know I should be able to figure this out, but the nested logic is not super easy to follow. Can we put in a comment a 'truth table' or something to show the cases we're evaluating here?

[jzuhone]

Done

[jzuhone]

pre-commit.ci autofix

[jzuhone]

I think this should be ready to go now.