refactor(api): differentiate between positional and keyword arguments in most expression APIs

[cpcloud]

Closes #9125.

[cpcloud]

Once this gets reviewed, I will give the individual commits a BREAKING CHANGE section so that the individual changes/recommendations show up in the 10.0 release notes.

[gforsyth]

Before I go through the rest of these -- I thought I should check and make sure I'm not missing something -- do we not want to standardize the read_* methods with forced positional and keyword args?

[cpcloud]

do we not want to standardize the read_* methods with forced positional and keyword args?

I think ultimately we do, which is why I ended up grinding some of them out earlier this morning.

[gforsyth]

CI failures notwithstanding, this looks good to me. There's some question remaining about whether we always, e.g. say path as the arg name to read_* functions regardless of whether it takes a list of paths, or if we want the distinction. But we don't have to handle that here, since making it positional-only means we can change it without breaking anyone.

[cpcloud]

Ok, well I decided to do the backends as well to the extent possible, might as well avoid 11.0 for as long as possible by doing some grunt work up front.

[cpcloud]

Now, the gruntiest part: creating a sequence of commits that will help users understand the breakage 😬