Implement broadcast arrays

[djpugh]

• Improves Numpy Supported functions #981
• Executed pre-commit run --all-files with no errors
• The change is fully covered by automated unit tests
• Documented in docs/ as appropriate
• Added an entry to the CHANGES file

[MichaelTiemannOSC]

I think I need this. I want to be able to broadcast a quantity to a dataframe thusly:

x['foo'] = Q_(1.0, ureg('m'))

Presently I get this error message:

TypeError: object of type 'int' has no len()

If I manually create a PintArray based on the length of X and do some extra work I can cause my scalar quantity to become part of the X dataframe. But a broadcast feature would be wonderful.

[djpugh]

@hgrecco - the previous tests failed on install of uncertainties==3.0.1

[keewis]

TypeError: object of type 'int' has no len()

Note that you can work around this by setting ureg.force_ndarray_like = True or ureg.force_ndarray = True

[djpugh]

@hgrecco this seems to be passing and ready to merge - any other comments?

[andrewgsavage]

@hgrecco this seems to be passing and ready to merge - any other comments?

could you change the line in CHANGES to be for the newest version?

[codspeed-hq]

CodSpeed Performance Report

Merging #1410 will not alter performance

_{Comparing djpugh:implement_broadcast_arrays (87c18c7) with master (24dd237)}

Summary

✅ 437 untouched benchmarks

[andrewgsavage]

I implemented broadcast_arrays after this PR (didn't see this one, my bad)
Looks like the implementation in this PR doesn't work with the test case I made, so this will need looking into

[djpugh]

@andrewgsavage, I've not had to use this functionality in a while, although I'm now working on something similar to last time I hit this error (and implemented this). Happy to look at it again if it's still an issue