Coq 8.6

These are the changes necessary to make this compatible with Coq 8.6. Most of the changes are fine (and of the category "no idea why this worked in 8.5" or "the statement of a lemma changed in the 8.6 libs"), except for the horribleness in perm_incl.v. I played around a little but found no good way to restore the term into the state Coq 8.5 put it in -- and the goals I end up in otherwise (if I just remove the change) look fairly unsolvable.

Cc @jjourdan @robbertkrebbers

theories/lang/races.v

@@ -201,7 +201,7 @@ Proof.
 
   destruct Ha1 as [[]|[]| | |], Ha2 as [[]|[]| | |]=>//=; simpl in *;
     repeat match goal with
-    | H : _ = Na1Ord → _ |- _ => specialize (H (reflexivity Na1Ord)) || clear H
+    | H : _ = Na1Ord → _ |- _ => specialize (H (eq_refl Na1Ord)) || clear H
     | H : False |- _ => destruct H
     | H : ∃ _, _ |- _ => destruct H
     end;