The benchmark help output says R is the default RVM, but R-bytecode is the actual default

The rbench.py help output says R is the default RVM, but R-bytecode is the actual default.

The bug is clearly caused by the inconsistency between these two lines in `rbench.py`:

```
parser.add_argument('--rvm', default=rvms[0],
    help='R VM used for the benchmark. Defined in rbench.cfg. Default is '+rvms[1])
```
