Launch pothos-benchmark toolkit

[guruofquality]

With the new feature of JSON topologies and JSON stats reporting, and nicely bundled up into the PothoUtil executable: https://github.com/pothosware/pothos-library/wiki/PothosUtilGuide#running-a-json-topology ... it should be possible to make a python-based benchmark suite that executes a variety of benchmarks, collecting the data, and plotting it in matplotlib to create nice comparison barcharts.

This helps to establish a very encompasing baseline to determine how future features and performance work affects anything and everything. Most of the benchmarks will probably use passthrough blocks like router, gateway, black hole, infinite source... so that scheduler overhead is the dominating factor in the benchmarks.

Interesting things to vary:

• machines, arm, x86, NUMA nodes
• thread pool vs thread per actor
• waiting on CVs vs spining
• extreme topologies - linear-chain, massive fan-in, fan-out, hyper-mesh

Close this issue when the benchmark toolkit is created and in a mostly useful state.