I'm iffy about this tag, but I think there's a workable core here.
Part of it is that I need to be able to see what tags belong on a question by looking at the question, not at the answers. That means when this part comes up...
[...] and that can't always be cleanly answered with a CR reference.
... that means I don't necessarily know whether a question is a templating question or not until it's answered. Tags don't handle this Schrödinger situation very well since part of their purpose is to connect questions with experts who will answer it, so if I only know how to tag it after it's already answered, something is wrong.
Part of it is also, what is and isn't a templating question? One of your examples isn't a templating question, it's a legend rule question: What happens if the legend rule doesn't apply to only one of a pair of legendary creatures with the same name?. Maybe that's a bad example. But to some extent “templating” means “the text on the card”, and all questions about specific cards come down to the text on the card. I suspect this would be a little too broad of a tag if we're applying it in such a way this example might be included.
I see where you're getting at with two of your examples:
... since these are about templating qua templating. They're not “how does this card work?” but instead “why does the card say these words specifically?” or “does my card need to say these exact words in order to work the way I intend?” These are drawing lessons about how people get the intended functionality of a card into that card's rules text within the bounds of the CR, which is what templating is really about from a MTG R&D department point of view.
I think this might be the core of the tag that you're imagining, and I suspect these aren't the only questions we have that are like this. This will be workable since we'll be able to see what it is based on the question itself, and we'll have clearer lines to draw.