MathMLElement doesn't reflect attributes for more specific elements

[Jamesernator]

Generally in HTML when an element has attributes other than the global attributes it has a more specific interface allowing those attributes to be reflected, for example <link> has HTMLLinkElement, or h1-h6 have HTMLHeadingElement.

Currently MathML core only specifies a single MathMLElement which can only reflect mathml global attributes, despite there existing multiple elements with additional attributes:

• <mo>
• <mspace>
• <mfrac>
• <mpadded>
• <munderover>
• <mtd>

I would propose that those elements should get their own interfaces with the appropriate attribute reflection.

As an example for <mo> this would allow setting el.form = "infix" rather than using .setAttribute.

[bkardell]

I'm not opposed to this, but I'm tagging this for consideration in L2 as there are already some questions around this for things like how do do links and what IDL would it require/what limits does it impose for L2. It's already been a lift to get MathMLElement and an agreeable definition of what that is that is non-problematic (previously it all were just Element), not all HTML attributes reflect and choices here have other impacts - I think it's worth considering, but not blocking a good L1.

If you disagree with that lmk.

[fred-wang]

Agree with Brian. Also same comment as in #166 (comment) : I'd really like we are clear the set is stable and that we don't introduce a lot more attributes in the future...

Additionally, I wonder whether one could write a polyfill to help developers?