Two observations:
- prependers where both arguments are null are common in typical expressions - caching these is profitable
- prepender argument ordering is unfortunate: String prefix, <arg>, String suffix. This means we have to do two back-to-back insertArguments rather than one

Additionally we should add a bridge to the String mix function in StringConcatHelper to avoid method handle invokes in the bootstrap path.

Patch: http://cr.openjdk.java.net/~redestad/8245756/open.00/

All together I see a substantial improvement on a number of concatenation bootstrap stress tests. For example, running: http://cr.openjdk.java.net/~redestad/scratch/erase_scf_types.00/ObjStringCombos.java with perf stat -r .. shows these numbers before:

     4,257,229,918 instructions # 1.10 insn per cycle ( +- 0.84% )
       847,230,764 branches # 504.826 M/sec ( +- 0.87% )
        34,600,787 branch-misses # 4.08% of all branches ( +- 0.91% )
       0.319545331 seconds time elapsed ( +- 1.18% )

And after:

     3,692,867,129 instructions # 1.10 insn per cycle ( +- 0.73% )
       734,054,515 branches # 498.753 M/sec ( +- 0.74% )
        30,699,692 branch-misses # 4.18% of all branches ( +- 0.75% )
       0.295842958 seconds time elapsed ( +- 1.10% )