New definition of contractive.

The current notion of `Contractive` does not allow one to deal with functions with multiple arguments, for example, binary functions that are contractive in both arguments (like `lft_vs` in lambdarust), or binary functions that are contractive in one of their arguments.

To that end, I propose I reformulate the notion of `Contractive` so that we can express being contractive using a `Proper`. The new definition is:

    Definition dist_later {A : ofeT} (n : nat) (x y : A) : Prop :=
      match n with 0 => True | S n => x ≡{n}≡ y end.
    Notation Contractive f := (∀ n, Proper (dist_later n ==> dist n) f).

Also, it turns out that using this definition we can implement a `solve_contractive` tactic in the same way as the `solve_proper` tactic.

Unfortunately, the new tactic does not quite work for the weakest precondition connective in Iris because the proof involves induction, and the induction hypothesis does not quite fit into the new `solve_contractive` tactic.

See merge request !32

Using this new definition we can express being contractive using a
Proper. This has the following advantages:

- It makes it easier to state that a function with multiple arguments
  is contractive (in all or some arguments).
- A solve_contractive tactic can be implemented by extending the
  solve_proper tactic.

This removes Ralf's hack of using later_car, which is not function in the logic.

Thanks to Aleš for suggesting this.

Also, higher cost for [elim_modal_bupd_fupd], so that it is not taken in place of [elim_modal_fupd_fupd] in spec patterns.

When having H : ▷ (P -∗ Q) and H2 : ▷ P, iSpecialize ("H" with "H2")
distributes the later over the wand.

The rewrite auth_validN_eq was not performed in the hypothesis. It
used to work in 8.5 because of magic.

Also, use explicit unfolding lemmas for auth_valid and auth_validN.
The `Arguments valid _ _ !_ /` hack did not really work when one
has to deal with the valid instance of the cmra, which underneath also
includes a `cmra_valid`. Declaring a similar Arguments for `cmra_valid`
is a bad idea, it will also end up unfold stuff for the exclusive and
option CMRA.

This fixes issue #46.

Proof was done by Hai & me

This fixes issue #44.