RFR (S): Inline and optimize matrix operations for partial GC

[Aleksey Shipilev]

Hi,

Trivial things:
  http://cr.openjdk.java.net/~shade/shenandoah/partial-prepare-perf-1/webrev.01/

Improve performance quite a bit:

Partial GC (net)  =    24.58 s (a =     5858 us) (n =  4197)
   (lvls, us =     1016,     4785,     5840,     6914,    11481)

Partial GC (net)  =    16.41 s (a =     3736 us) (n =  4392)
   (lvls, us =      832,     3320,     3809,     4062,    62067)

Testing: hotspot_gc_shenandoah, benchmarks

-Aleksey