fix: use isinstance() instead of hasattr()

[rich-iannone]

In #348 we wanted to ensure that the signature of tab_spanner() has been updated with types that indicate md() and html() could be used in the label= argument. This was done in a previous PR.

A second requirement was update the logic to use isinstance() checks rather than hasattr(). That change was made in this PR.

Fixes: #348

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 86.37%. Comparing base (11660a6) to head (fe684d4).
Report is 4 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #434   +/-   ##
=======================================
  Coverage   86.37%   86.37%           
=======================================
  Files          42       42           
  Lines        4683     4683           
=======================================
  Hits         4045     4045           
  Misses        638      638           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[machow]

Can you show that #348 is resolved? (either by pointing to tests or creating examples)

• If md() is passed is it handled as expected? (e.g. does the markdown get rendered)
• If html() is passed is it handled as expected? (e.g. does the html get rendered)
• If text is passed that looks like html, it should not get rendered as html (if this is happening, even if we close issue 348 we should track this bug; but also we may have an issue open for this already?)

I'm surprised that we could simply grab the text our so that md() and html() is a no-op

[machow]

I got this so wrong 😭 -- these changes are for handling id, which is the unique name for referring to the spanner. But label display is handled later on in the code (https://github.com/posit-dev/great-tables/pull/434/files#diff-b4c74a432c9ffd70ea3c383684233bfdea5f3bac27dacab684be0c3a4e9cffb5L190).