Introduce `set_bind` and associated lemmas + set_bind theory: revise setoid rewriting

This is @dfrumin's !383 (closed) (with authorship preserved) + some tweaks to setoid rewriting:

• strengthen set_bind_subset to have the same arity
• add missing Params instance to prevent slow failures in setoid rewriting
• Proper instances can be proven before set_bind_ext via set_solver.
• set_bind_ext is then provable by set_solver directly.

I originally sent dfrumin/coq-stdpp!1 (closed) to be added into !383 (closed), but maybe it's easier with this MR? @robbertkrebbers up to you.