upgrade hashbrown

[matklad]

The primary goal is to remove old hashbrown from nearcore.

I considered making a larger change and actually hiding hashbrown altogether, but, on ballance, this seems like the best first step. If we want to break the API, there's quite a few things we might want to do #51

[austinabell]

I think this case is pretty impossible to hit for users where this is a breaking change and anyone building a contract should already have a lockfile. I can't personally think of a reasonable way this would cause any issue to be a patch release