Interaction between height and baseline alignment

Follow-up of #75

I've worked on the Firefox implementation of width/height, tried to follow what MathML Core / Chromium do and I'm adding missing test coverage.

I just uploaded a testcase for various situations: https://bug1916988.bmoattachments.org/attachment.cgi?id=9425482

One behavior I'm not sure is what happens when we specify a CSS height that is smaller than the line-ascent. It seems Chrome sets the corresponding height and does not clamp the line-ascent, so baseline alignment with other siblings stay the same. I think that makes sense to me and that's what I'm implementing in Firefox, but we might need some extra text is the spec to clarify.

Related to that, when we specify a height for the "content box" that is different from the one of the "math content box", then MathML Core (and Chromium, in what I'm implementing in Firefox and in WPT tests I'm adding) says to align the top edges of these boxes. But maybe what is more natural is to tweak the line-ascent/line-descent evenly so that the baseline is unchanged.

Start by reading the follow-up context in issue #75 and reviewing the uploaded testcase. Compare the described height, line-ascent, line-descent, and baseline-alignment cases with MathML Core and the related WPT coverage. The work is done when the intended behavior is decided and the specification text and tests clearly reflect it.