Packets on one path must not adjust values for a different path

[erickinnear]

The intent here was that either (a) you only ever have a singe CC context and you reset it for each path, or (b) you have a CC context for each path, in which case each context is strongly tied to a path and you don't need to reset anything (except when it's been a while and the previous values may no longer be valid, etc.). We require that everyone does at least (a), while not preventing folks who want to do extra work for a potential performance gain from doing (b).

Packets can contribute to the CC/RTT estimation for the old path if it's still around and you're willing to maintain that state, which is why this (mistakenly) was a SHOULD.

Closes #2909.

[martinthomson]

This looks good. I'm calling it editorial. However, it changes normative language, so I want an opinion on that from at least @ianswett or @janaiyengar before merging (other opinions are also highly valued).

[janaiyengar]

+1 that we should require this behavior. We know that implementers will try out tricks here if they want to, and there's no enforcement to stop that from happening.

[erickinnear]

I agree that this seems generally editorial, but either way I think this is ready to go?