RFR: Shenandoah region/set iterators should not allow copying

[Aleksey Shipilev]

http://cr.openjdk.java.net/~shade/shenandoah/iterator-copy-mess/webrev.01/

We have a major performance regression like this:

[7.816s][info][gc] GC(8) Pause Final Update Refs 40.418ms

The cause is "Bitmap based ShHeapRegionSet" change [1] that messed up iterator copying. The handover
to final-update-refs walked all the regions *again* during the pause, because the iterator was
passed by value, and got index=0. I fixed both new iterators just in case.

Testing: hotspot_gc_shenandoah

[1] http://hg.openjdk.java.net/shenandoah/jdk/rev/a4691713454d