BUG - Fix a11y tests call

[trallard]

Closes #1319

I tested this locally with nox -s a11y, and the accessibility tests are being called correctly.
They are finding several " expected " issues; as we know, there are several accessibility violations here.

If the CI is happy, this should be ready for review and merge.

EDIT:

• The tests ran on CI see https://github.com/pydata/pydata-sphinx-theme/actions/runs/4931377199/jobs/8813353060?pr=1320
• Of course, now the workflow is marked as failed. I wonder if it is worth marking the a11y tests step as continue-on-error while we work on fixing these accessibility issues (cc @drammock)

[drammock]

I pushed commits to fix a comment typo and to add continue-on-error (which I think is a good approach, thanks for suggesting it).

In the long run I'd like to move toward:

• having some CI jobs required for merge (and the a11y test not required)
• once we get the a11y test green, making it also required for merge

but for today let's be satisfied that the tests actually run now :)

[trallard]

Fabulous many thanks. That approach for required workflows sounds good.
Been swamped but I should be able to send another PR soon to improve the reporting of accessibility tests and coverage.

[drammock]

Proof that it worked:
https://github.com/pydata/pydata-sphinx-theme/actions/runs/4938848103/jobs/8829045824?pr=1320#step:11:1411

It turns out that continue-on-error has the unfortunate side-effect of marking the job as successful (!) so we'll need to click through to the CI log to actually see whether there were a11y test failures. Apparently an alternative is to mark any subsequent steps in the job as if: always() --- this lets subsequent steps run while still marking the failed step with a red X. However:

• the a11y test is currently the last step in its job anyway so we probably don't need either continue-on-error or if: always()
• the a11y test should probably be its own job rather than a step at the end of the main pytest job.

I'll split this off as a separate issue to remind us to maybe reorg the workflows file in future.

(also I'm ignoring the failed profile job because GitHub is just having a bad day (keeps failing with "internal error"))