[pre-commit.ci] pre-commit autoupdate

[pre-commit-ci]

updates:

• github.com/adamchainz/blacken-docs: 1.13.0 → 1.14.0
• github.com/charliermarsh/ruff-pre-commit: v0.0.260 → v0.0.275

[neutrinoceros]

Looks like ruff upgrading string interpolation style leads to syntax errors when said interpolated strings are processed through eval, but only on Python 3.8 and 3.9;
It may be the case that unyt is doing something fragile with string interpolation, or that string interpolation rules are slightly different in Python 3.10 and ruff wrongly assumes the change is backward compatible for all Python versions.
I don't have the time necessary to sort it out at the moment but I intend to come back to this later.

[neutrinoceros]

The issue is that formatting methods are not equivalent for unyt_quantity objects

>>> import unyt as un
>>> a=1.0*un.cm
>>> print("%f" % a)
1.000000
>>> print("{:f}".format(a))
1.000000 cm

Note that this isn't dependent on the Python version we just happen to not hit the broken code on Python>=3.10 because many tests still require nose. Let me see if this can be fixed in unyt

[neutrinoceros]

I currently do not understand how to change the behaviour of "%f" % a (overriding __mod__ and __rmod__ operators on unyt_array doesn't work), and even if it's possible, it would resolve the issue here, so let's try to use views instead.

[neutrinoceros]

I suspect this is a breaking change too and that this function isn't fully tested

[neutrinoceros]

checked. It's not breaking because RadMC3DSource.position is a list of floats, without explicit units.