RFR(M): 8223051: support loops with long (64b) trip counts

[Roland Westrelin]

> So can we arrange to run it more than once, by setting the inner trip
> count to be smaller?  I’m afraid the optimizer could detect a one-trip loop
> and take it apart (in a later pass), and then the goal of the stress test
> won’t be achieved.

Sure. Here is a new patch that should take all of your comments into
account:

http://cr.openjdk.java.net/~roland/8223051/webrev.01/

I split the change in 2. The counted loop refactoring code is out for
review under 8244504 (I just sent it out for review). Your review of
that other one is very much welcome.

Roland.