ACTION-1744: Implement edits proposed in issue-641, raise to group if any subsantive changes come out

It turns out that name-from was turned to "n/a" as part of action-1440.
https://github.com/w3c/aria/commit/ad4542315

Regarding math and definition: Do we really want to allow name from contents??

As a reminder, a while back we removed name from contents for listitem. The reasons for doing so were first and foremost the fact that there may be a bunch of descendants (such as a nested list). I can see that being the case both for math (see the MathML torture test) and for definition (imagine the 20-volume OED as a web app). Another reason is that what an AT likely wants to present for listitem (and I would argue definition and math) is the text under the element; not a name. And it might want to present that text in smaller chunks (e.g. during caret navigation); not all at once. Lastly, if an author really has a need to override this behavior, listitem (and definition and math) already allow name from author. Long way of saying that I don't think we want to make this change.

I'm less certain about image. But given a sufficiently complicated image (e.g. an organizational chart), I question both the desirability of doing it at all. If we were to do it, having the user agent programmatically do it raises new questions: Should it be DOM order? Visual order? If the latter, top-down or left-right? Do we need to take locale into consideration? The more I think about it, the more I think this seems like a not-good idea....

We should not do this.

1. MathMl markup should not be not be stringified and we don't have name compuations on MathML.
2. definitions can be HUGE. we don't want that in a name that is supposed to be short

Closing. As indicated in the notes, "We should not do this."