Consider removing either mdc-elevation-transition or mdc-elevation-transition-rule

[kfranqueiro]

We currently have a function and a mixin that do nearly the same thing:

• mdc-elevation-transition-rule, which returns the right-hand-side for a transition property value
• mdc-elevation-transition, which emits that transition rule plus adds will-change: box-shadow

Currently we use the latter in 3 or so places ourselves, and only use the former within the latter.

However, the automatic/forced inclusion of will-change: box-shadow contributed to bug #1665, which makes me wonder if including it by default might not be the best idea. Either way, we do expose a CSS class which applies both transition and will-change, and we could keep that unchanged. So my vote would be to remove mdc-elevation-transition and keep ...-rule.

Another option would be to nix the function and remove will-change from the mixin, thus making the mixin equivalent to what the function currently does, but emitting the style for you.

Either way, I'm not sure there's a reason for both of these things to exist when they're so similar.

[kfranqueiro]

Another argument for favoring the -rule function over the mixin is it enables you to combine the value it returns with other transition properties within a single rule. (h/t ErraticFox on discord for bringing up this use case.)

I'm going to go about removing the mixin and using the -rule function everywhere, so that I can safely move the ripple will-change property back to the surface and off of the pseudo-elements (which will resolve #1753 and #1854 without regressing #1665).