future proofs: Future.fork_pri 1 minimizes queue length and pending promises
-- slight improvement of throughput, at the cost of a bit of parallelism;

--- a/src/Pure/Isar/toplevel.ML	Tue Dec 16 16:25:20 2008 +0100
+++ b/src/Pure/Isar/toplevel.ML	Tue Dec 16 18:04:16 2008 +0100
@@ -718,7 +718,7 @@
 
         val future_proof = Proof.future_proof
           (fn prf =>
-            Future.fork_pri ~1 (fn () =>
+            Future.fork_pri 1 (fn () =>
               let val (states, State (result_node, _)) =
                 (case st' of State (SOME (Proof (_, (_, orig_gthy)), exit), prev)
                   => State (SOME (Proof (ProofNode.init prf, (finish, orig_gthy)), exit), prev))

--- a/src/Pure/goal.ML	Tue Dec 16 16:25:20 2008 +0100
+++ b/src/Pure/goal.ML	Tue Dec 16 18:04:16 2008 +0100
@@ -179,7 +179,7 @@
     val res =
       if immediate orelse #maxidx (Thm.rep_cterm stmt) >= 0 orelse not (Future.enabled ())
       then result ()
-      else future_result ctxt' (Future.fork_pri ~1 result) (Thm.term_of stmt);
+      else future_result ctxt' (Future.fork_pri 1 result) (Thm.term_of stmt);
   in
     Conjunction.elim_balanced (length props) res
     |> map (Assumption.export false ctxt' ctxt)