Update calver dependencies to use inequalities

[sjperkins]

• Tests added / passed

  $ py.test -v -s daskms/tests

  If the pep8 tests fail, the quickest way to correct
  this is to run autopep8 and then flake8 and
  pycodestyle to fix the remaining issues.

  $ pip install -U autopep8 flake8 pycodestyle
  $ autopep8 -r -i daskms
  $ flake8 daskms
  $ pycodestyle daskms
  

• Fully documented, including HISTORY.rst for all changes
  and one of the docs/*-api.rst files for new API

  To build the docs locally:

  pip install -r requirements.readthedocs.txt
  cd docs
  READTHEDOCS=True make html
  

[sjperkins]

Test cases pass

[sjperkins]

This PR changes dask-ms to install 2024 versions of calver'd software, mostly because they're specified using poetry's caret dependency scheme

This is because a ^2023.X.Y specifier results in a >= 2023.X.Y, < 2024.0.0 dependency range.

OTOH it might be possible to just use inequality requirements and keep the range lower.

But I think we'll need stricter requirements going forward due to task keys on the distributed scheduler changing from strings to tuples, as well as the upcoming integration of dask-expr into dask.

[JSKenyon]

One comment is that restricting the versioning to a specific year (calver) makes for weirdly quantised dependencies. I think that an inequality constraint is fine and means that things can change more smoothly. I do agree that there may be trouble on the horizon due to upstream churn.

Edit: My approach going forward is to mandate poetry installation for end-users.

[sjperkins]

OK, I've relaxed the calver dependencies to inequality constraints. In combination with the python version, I think poetry will make sensible choices here.