typo fix

[estelle]

<string> was considered a tag, so switched to <dfn><<string>></dfn> so it would be visible

[dontcallmedom]

Marked as non substantive for IPR from ash-nazg.

[frivoal]

The original was definitely wrong, good catch, and thanks for the patch.

As to the correction, I wonder what's better. What you did is one option, but we're not really defining the type <<string>> here, so maybe just marking it up as <<string>> (so that it links to css-values-3 would be more appropriate than <dfn><<string>></dfn>.

I know there's a handful of other places in other specs that follow the pattern use used, but it feels off to me, and I think they too should be corrected.

If we really want to enable cross-linking between the grammar and the definition, I think what would be more appropriate is to name this non terminal. Something like this:

<dt><dfn>path()</dfn> =
	path( [<<fill-rule>>,]? <<path-string>> )
</dt>
[...]
<dfn><<path-string>></dfn> is a <<string>> that represents an
<a href="https://www.w3.org/TR/SVG11/paths.html#PathData">SVG Path data string</a>.

I'm approving your PR nonetheless, because you're fixing a clear bug, while what I'm saying feels somewhat subjective. So feel free to merge if you disagree, and at least we'll get the bug fixed, but I think it'd be better either without a dfn, or with a named non-terminal.

[tabatkins]

The correct markup for things like this is just to mark it as for the appropriate thing. That's used regularly in property definitions. I'll edit the PR and merge it.