REG: recent regression in field unit detection

[neutrinoceros]

Bug report

Bug summary

Found a regression on the main branch that was discovered via a test added in #3849

this problem bisects to 60e9414 (#3921)

Code for reproduction

Here's a minimal reproducer, using dimensionless test arrays (noise1 and noise3)

from yt.testing import add_noise_fields, fake_random_ds
from yt.visualization.api import PhasePlot

ds = fake_random_ds(16)
add_noise_fields(ds)

my_sphere = ds.sphere("c", 1)
p = PhasePlot(
    my_sphere,
    ("gas", "noise1"),
    ("gas", "noise3"),
    [("gas", "density")],
)

Actual outcome

yt : [INFO     ] 2022-06-07 19:15:37,464 Parameters: current_time              = 0.0
yt : [INFO     ] 2022-06-07 19:15:37,464 Parameters: domain_dimensions         = [16 16 16]
yt : [INFO     ] 2022-06-07 19:15:37,465 Parameters: domain_left_edge          = [0. 0. 0.]
yt : [INFO     ] 2022-06-07 19:15:37,465 Parameters: domain_right_edge         = [1. 1. 1.]
yt : [INFO     ] 2022-06-07 19:15:37,466 Parameters: cosmological_simulation   = 0
yt : [WARNING  ] 2022-06-07 19:15:37,909 Field ('gas', 'noise1') was added without specifying units or dimensions, auto setting units to 
yt : [WARNING  ] 2022-06-07 19:15:37,909 Field ('gas', 'noise3') was added without specifying units or dimensions, auto setting units to 
Traceback (most recent call last):
  File "/Users/robcleme/dev/yt-project/yt/t_reg_phase.py", line 8, in <module>
    p = PhasePlot(
  File "/Users/robcleme/dev/yt-project/yt/yt/visualization/profile_plotter.py", line 971, in __init__
    profile = create_profile(
  File "/Users/robcleme/dev/yt-project/yt/yt/data_objects/profiles.py", line 1454, in create_profile
    obj = cls(*args, **kwargs)
  File "/Users/robcleme/dev/yt-project/yt/yt/data_objects/profiles.py", line 733, in __init__
    x_min, x_max = _sanitize_min_max_units(
  File "/Users/robcleme/dev/yt-project/yt/yt/data_objects/profiles.py", line 35, in _sanitize_min_max_units
    rmin = amin.in_units(finfo.output_units)
  File "/Users/robcleme/.pyenv/versions/yt-dev/lib/python3.10/site-packages/unyt/array.py", line 821, in in_units
    units = _sanitize_units_convert(units, self.units.registry)
  File "/Users/robcleme/.pyenv/versions/yt-dev/lib/python3.10/site-packages/unyt/array.py", line 259, in _sanitize_units_convert
    unit = Unit(possible_units, registry=registry)
  File "/Users/robcleme/.pyenv/versions/yt-dev/lib/python3.10/site-packages/unyt/unit_object.py", line 227, in __new__
    raise UnitParseError(
unyt.exceptions.UnitParseError: Unit representation must be a string or sympy Expr. 'None' has type '<class 'NoneType'>'.

[neutrinoceros]

So I see two reasons why this is failing:

• noise test fields are not defined with explicit units (they are meant to be dimensionless). This should be fixed since RFC: a more robust plot norm/colorbar API #3849 makes not passing a unit equivalent to passing units="auto", rather than units=" (dimensionless).
• at initialisation, profiles retrieve units from data_source.ds.field_info before calling YTSelectionContainer._generate_fields, so it doesn't work with ENH: improve error messages when add_field receives a malformed function argument #3921

I think this is a grave oversight from #3921, and the best course of action right now is to revert it.

[neutrinoceros]

at initialisation, profiles retrieve units from data_source.ds.field_info before calling YTSelectionContainer._generate_fields, so it doesn't work with #3921

After more careful checking, this turns out to be wrong, hence invalidating my previous conclusion

[neutrinoceros]

I think I found the proper fix in #3962