Introduce JMH Benchmarks

[lloydmeta]

Part of #33 involves adding JMH-backed benchmarks so we can keep track of performance. This PR aims to introduce some.

To run benchmarks:

$ ./run-benchmarks

Changes:

• Add a subproject for running benchmarks called 'benchmarks'
• Add an SBT Plugin dependency on SBT-JMH
• Add tests for the blocking actions exposed by Shade.Memcached
• Add a script for running the benchmarks

[coveralls]

Coverage remained the same at 85.892% when pulling 149aff0b2f95334a14081dd9ef4b64867eb3ede3 on lloydmeta:feature/benchmarking into 90fbbaeadb2cebab1e789ba5a8a60434800b55ea on alexandru:master.