(which are useful in Amin's logical relations development).

This generalization is surprisingly easy in Iris 3.0, so I could not
resist not doing it :).

In particular, remove ectx_language.atomic, because it seems unused
expect for a smart constructor for language.atomic.

* Values are considered as atomic expressions (this does not hurt, and
  this makes the proofs of atomicity simpler).

I also reverted 7952bca4 since there is no need for atomic to be a
boolean predicate anymore. Moreover, I introduced a hint database
fsaV for solving side-conditions related to FSAs, in particular,
side-conditions related to expressions being atomic.

and show that this is an instance of evaluation contexts

Just because we don't have such comments for other files, doesn't mean we can't
start applying some best practices.

This required a new ectx axiom: Positivity of evaluation contexts. This axiom was
also present in the old Iris 1.1 development, back when it still derived lifting
axioms for ectx languages.