introduce notation to provide "pres" existential witnesses by giving the missing proof. (This is a natural candidate for more ltac magic, to solve these obligations automatically... actually, why can't eauto do that?)

Simplified adv, defining it with ownL.

Proof of concept for a friendly interface that (if it works) lets the user
set up an invariant and prove view shifts and atomic triples for primitive
reductions, rather than work in the model. (It should work, but I have to
merge my two proofs to make sure.)

add first version of RA (resource algebra): validity is a decidable predicate, rather than hard-coded into the type structure

create iris_meta file for meta-theorems (adequacy, lifting lemmas). formulate two of our four lifting lemmas.

improve n[] notation for nonexpansive maps: the proof of Proper is no longer required, it can be derived from nonexpansiveness

This reverts commit 608fe86e22b912d9d591cd2d0c4e2943b1abe6ce.

protocols where I want to prove something called robust safety.
Ironically, to even state robust safety requires Hoare triples that
don't imply safety. So Iris supports both {P} e {Q} (implying safety)
and [P] e [Q] (not). I'll add a rule for forgetting about safety:

	{P} e {Q}
	— Unsafe
	[P] e [Q]

some time soon.

Aside: I'm an SSReflect weenie and know next to nothing about the
usual Coq tactics. My proof script changes likely reflect that fact.