Refactor close-matching when reading toml config

[maciejb]

Describe your changes

This PR refactors sql.util.get_user_configs() to more specifically target only the config section [tool.jupysql.SqlMagic].

When searching for "jupysql", a hint (instead of an error) will be displayed only if there is exact case-insensitive match (such as "jupySql"). We are making this stricter, because we cannot anticipate the naming of other tools in the general tool namespace.

A close_match is still used for "SqlMagic", so we detect things like "SqlMagics". If found, such things will continue to throw an error with a hint, because we are now in the jupysql namespace.

Because the logic differs slightly at each level, I revised the structure to be a direct linear derival of the key we need, rather than the more generic toml loop search implementation. This enhances readability over the alternative of adding if/else logic within the loop to vary the logic as we go deeper down the tree.

Issue number

Closes #975
Closes #969 -- this is fixed by the return None, None at the end. Although not original target of this PR, it wasn't possible to just leave this broken

Checklist before requesting a review

• Performed a self-review of my code
• Formatted my code with pkgmt format
• Added tests (when necessary).
• Added docstring documentation and update the changelog (when needed)

---

📚 Documentation preview 📚: https://jupysql--976.org.readthedocs.build/en/976/

[edublancas]

@neelasha23 @bryannho please review

[bryannho]

Local testing works fine and code looks good to me

[bryannho]

Whoops, functionality is fine but before we can merge you'll need to add an entry to CHANGELOG.MD or add the no-changelog label and then re-run the CI!

Additional review below

[neelasha23]

Please add a Changelog entry and ensure all the tests are passing @maciejb

[bryannho]

@maciejb is this ready for review? don't worry about the check-for-broken-links test

[maciejb]

The one failing CI test seems to be a prior issue unrelated to my changes, a false-positive on a broken link in doc/integrations/mssql.ipynb to medium.com. I believe that is something which should be addressed upstream (either add medium.com here or fake a user agent when making the request)

[maciejb]

@maciejb is this ready for review? don't worry about the check-for-broken-links test

@bryannho oops, just saw this after I traced back the check-for-broken-links test :)

Yes, ready for review, with exception of I don't know whether or not to address @neelasha23's comment here. I do think removing this test is the right to do (not xfail), but whether to replace it with a different test I'm not sure about.

[edublancas]

@maciejb let us know if you still have time to address the review!

[maciejb]

@maciejb let us know if you still have time to address the review!

@edublancas , thanks for the ping. Sorry, this turned from a simple "remove extraneous and buggy functionality" into something a little different. Got busy with work and lost track of this. I'll try to throw something together shortly.

[edublancas]

sounds good!

[maciejb]

Thanks all, I believe this is now ready to go!

[bryannho]

Manual testing works great, code looks good to me. Thanks for your contribution @maciejb !

[edublancas]

thanks for working on this @maciejb!