This seems to shorten type class search.

Also, slightly reorganize.

Thanks to Amin Timany for the suggestion.

Contrary to destruct_conj from Program.

It now turns setoid equalities into Leibniz equalities when possible,
and substitutes those.

In principle, we could now un-seal heap_mapsto, saved_prop_own etc., and mark them as "Typeclass Opaque", and ecancel would still work just as fast as it does now.
Thanks to Matthieu for pointing me to this unify feature.

Conflicts:
	prelude/tactics.v

This replaces f_equiv and solve_proper with our own, hopefully better, versions

It now traverses terms at most once, whereas the setoid_rewrite
approach was travering terms many times. Also, the tactic can now
be extended by defining type class instances.

Also, use "set_solver by tac" to specify a tactic.

And now the part that I forgot to commit.

In most cases there is a lot of duplicate proof search performed by
both naive_solver and eauto. Especially since naive_solver calls its
tactic (in the case of set_solver this used to be eauto) quite eagerly
this made it very slow.

Note that set_solver is this too slow and should be improved.