RFR: Purge some G1-related cruft in C2

[Aleksey Shipilev]

On 09/11/2018 10:13 PM, Roman Kennke wrote:
> This gets rid of Node::is_g1_marking_load() and Node::is_g1_marking_if()
> .. some fairly obvious leftovers from GC-interface-ification, and fixed
> related call-sites. This is closer to upstream and arguably/hopefully
> more correct.
> 
> http://cr.openjdk.java.net/~rkennke/purge-c2-cruft/webrev.00/

*) The change to is_shenandoah_marking_load in ShenandoahWriteBarrierNode::fix_ctrl looks like a
bugfix, please split it out in a separate changeset, so we are sure it is backported even if other
cleanups are not?

Otherwise looks good.

-Aleksey