Consolidate benchmark results

[pomadchin]

As the result of https://github.com/geotrellis/geotrellis-pointcloud/issues/34 we have a project with PDAL pipeline examples and GT DElaunay examples.

To summarize results we can put the code into the benchmark subproject and create JMH benchmarks.

As a report it would be helpful to have the following table:

    | dataset | resolution | delaunay generation in ms |
PDAL|         |            |                           |
GT  |         |            |                           |

That would describe us a side by sidereproducible comparison of PDAL vs GeoTrellis Delaunay performance.

Connects #34

[pomadchin]

To reproduce these benchmarks run:

$ ./sbt
> project benchmark
> jmh:run -i 10 -wi 5 -f1 -t1 .*DEMRasterSourceBench.*

	Dataset	Resolution	delaunay + rasterization, ms/op
PDAL (native rasterizer)	red-rocks	6.9375	888.188 ± 139.047
PDAL (JVM rasterizer)	red-rocks	6.9375	1802.555 ± 1048.380
GT	red-rocks	6.9375	2702.808 ± 188.939

I think these benchmarks are good enough to get the sense of a relative performance. At this point it makes not a lot of sense to test it on different resolution, it is expected that PDAL Delaunay + C++ rasterization would be faster.

[pomadchin]

The full report is here https://github.com/geotrellis/geotrellis-pointcloud/tree/master/benchmark