This makes detecting the context line a user clicks on slightly easier.  See new issue #33768 for discussion.  I am not sure now I would do this just for this reason, without trying with the label first.

Using a text also allows us to experiment with tagging the context keyword somehow, to see if it seems like an improvement.  Perhaps make it bold, or underlined, or the normal keyword color, or some variant thereof (grayed?).

So I am inclined to go ahead.  I did notice that most of the patch is straightforward substitution.  To me, this suggests that the code is pretty good.  Also, I think change 'label' (implentation) to 'context' (purpose) is a good change by itself.  I would keep this even if we reverted to using Label.