Some test changes for discussion:

(github.com)

 Actual:
 L127 --.rounded-1·>·.home-hero-signup·>·div·>·:nth-child(1)
 Expected:
 L127 ++.rounded-1·div·>·div:nth-child(1)

I would have expected this to go to .rounded-1·div·>·:nth-child(1) but the code is peppered with other rounded-1 classes so the original was relying on a div being the child of another div.
I'd be quite happy ditching this type of 'accidental' structural selection, but likely you have a different opinion?

I can't figure out how to automatically update the expected test output — I could do some more commits separating out the 'good' test result changes vs. the 'bad' ones

I'd be quite happy ditching this type of 'accidental' structural selection, but likely you have a different opinion?

Ok so it's possible to have the best of both worlds, i.e. only drop the tagName if it is redundant, but at the expense of checking both versions (with and without tagName). Would that be acceptable?