Reconsider implementation of Equatable and Hashable for HPACKHeaders

[Lukasa]

For historical reasons, HPACKHeaders does not pay attention to order or indexability when considering equality and hashability. We should consider whether this is appropriate for this type, or whether we should make the definition more strict.

[dnadoba]

We could also deprecate the conformance and add two views to it, ordered and unordered, and only conform the views to Equatable/Hashable.

[Lukasa]

I don't think that's really the right thing to do. We should commit the core type to one of those two positions, and where necessary we can always expose the other.