Refactor Gen.pick

[ashawley]

Code cherry-picked from #575

[ashawley]

I believe this was a refactor, and not a performance improvement, but the benchmark shows no change, regardless.

> bench/jmh:run -wi 3 -i 3 -f1 -t1 org.scalacheck.bench.GenBench
Benchmark                   (genSize)  (seedCount)  Mode  Cnt      Score       Error  Units
GenBench.arbitraryString          100          100  avgt    3   3418.104 ±   529.567  us/op
GenBench.asciiPrintableStr        100          100  avgt    3   2154.244 ±  1392.219  us/op
GenBench.const_                   100          100  avgt    3      4.276 ±     0.665  us/op
GenBench.double_                  100          100  avgt    3     12.827 ±     2.766  us/op
GenBench.dynamicFrequency         100          100  avgt    3   2299.389 ±  5622.303  us/op
GenBench.eitherIntInt             100          100  avgt    3     57.532 ±    12.060  us/op
GenBench.identifier               100          100  avgt    3   4628.981 ±   947.116  us/op
GenBench.int_                     100          100  avgt    3     16.736 ±     1.131  us/op
GenBench.listOfInt                100          100  avgt    3   1967.003 ±   138.548  us/op
GenBench.mapOfIntInt              100          100  avgt    3   4530.931 ±  1827.278  us/op
GenBench.oneOf                    100          100  avgt    3     24.985 ±    18.573  us/op
GenBench.optionInt                100          100  avgt    3     73.200 ±    37.056  us/op
GenBench.sequence                 100          100  avgt    3    299.786 ±   162.735  us/op
GenBench.staticFrequency          100          100  avgt    3   2098.952 ±  1148.657  us/op
GenBench.testFilter               100          100  avgt    3  25694.976 ± 60196.275  us/op
GenBench.zipIntInt                100          100  avgt    3     42.314 ±    26.204  us/op