Fix: CI: isort GitHub action doesn't report expected issues

[elliot-100]

I did misconfigure the black compatibility for the official isort action originally, but doesn't seem to be any documentation on how to do it correctly, and in any case it doesn't show diffs like it's supposed to (raised isort-action #90). So I've re-implemented using command-line interface.

It should really be configured to take config from pyproject.toml, but that's a possible enhancement; this is restoring expected functionality.

NB isort checks are expected to fail on this PR until existing codebase has corrected isort rule run on it. I think is clearest if I do this in another PR. Will do a dry run in my own fork first...

[elliot-100]

Hmmm, looks like I've hit a snag - I can't immediately fix the existing codebase, as the isort CI check is giving me more errors than when I run isort locally, even though commands appear to be equivalent.

Looks like local command is ignoring the example .py files which are outside the main package, CI including them; and there are maybe some unexpected subtleties when configuring isort using pyproject.toml. I'll have to come back to this in a day or two...

BTW I've switched to ruff format as a replacement for isort +black on other projects.

[Olen]

Did it work as expected?

[elliot-100]

No, in short. I've fixed files inside spond in #117 and the revised CI here in this PR doesn't flag any issues with them.

However there seem to be subtle differences in behaviour of isort on files in the root, outside tests and spond, when running the same isort command locally vs GitHub actions. It doesn't seem to be as simple as the former ignoring them, which was my naïve assumption as per notes in #117. I spotted this when pre-emptively testing on #107

[elliot-100]

One option would be to only enforce isort on tests and spond in CI for now and worry about other files (examples + manual 'test') later. Opinion on that?

[PS I do think this may not be feasible to fully fix, as isort seems to be semi-abandoned given long-running issues/questions about behaviour, doc gaps, etc]

[Olen]

I am always very pragmatic. Better get something that works partially, and fix the rest later (if needed) than having nothing at all.

[elliot-100]

OK, I'll do that, probably later today, after tidying up. Thanks