Tiling performance

[dgault]

Whats included in this PR:

• Java code for tiling tests
• Maven pom updates
• Benchmark scripts updates
• Jenkins scripts updates

What isn't included:

• analysis scripts

Tests:
Large Image
Series 1 - Bazla-14-100-brain - 2015-06-19 23.34.11.ndpi
Starting with a tile size of 30720x25856, then dividing the tile x and y by 2 until tile size reaches 16x16.

Medium Image
Series 2 - Bazla-14-100-brain - 2015-06-19 23.34.11.ndpi
Starting with a tile size of 7680x6464, then subtracting 64 from the tile x and y until tile size reaches 16x16.

Manual Tiling
tubhiswt_C0_TP0.ome.tif
Starting with a tile of size 512x512, then dividing the tile x and y by 2 until tile size reaches 16x16. Writes each tile individually.

Auto Tiling
tubhiswt_C0_TP0.ome.tif
Starting with a tile of size 512x512, then dividing the tile x and y by 2 until tile size reaches 16x16. Writes a full plane with the auto tiling.

Available parameters:
iterations
input
output
results
tileXStart
tileYStart
tileXEnd
tileYEnd
tileOperator (either / or -)
tileIncrement (how much to divide or subtract tile size by)
series
autoTile

[rleigh-codelibre]

As mentioned on the trello card I think here it would be useful to put the new logic into a separate batch file (e.g. tiling.bat), move the other logic into metadata.bat and pixeldata.bat, and then have jenkins-build.bat call tiling.bat. The common logic at the top of jenkins-build.bat can stay as it is--this is just splitting out the test invocation so we can select the tests we want to run for each benchmarking run; for tiling we don't need to run the other tests.

[rleigh-codelibre]

The same applies equally to the Unix shell script.

[rleigh-codelibre]

This section could all be replaced by conversion from CoreMetadata using MetadataTools. Not strictly necessary, but the helper will ensure the metadata contains the minimal necessary information.

[rleigh-codelibre]

Would this be better named planes, since it's a list of planes containing a list of tiles?

[rleigh-codelibre]

Would this be better named tiles, since it's a list of tiles within a plane?

[sbesson]

After pulling the latest Docker image including this PR (from http://hub.docker.com/r/snoopycrimecop/ome-files-performance) and executing, the benchmarking went through except for the tiling execution because of the file name

[sbesson@ome-c6220-1-4 ~]$ docker run --rm -it -v /scratch/tmp/data:/data snoopycrimecop/ome-files-performance:master_merge_trigger 
+ datapath=/data
+ binpath=/install/bin
+ resultpath=/data/results
+ outpath=/data/out
+ iterations=1
+ mkdir -p /data/results
...
+ for test in neff-histopathology tubhiswt
+ input=unknown
+ case "$test" in
+ input=/data/ndpi/neff-histopathology/Bazla-14-100-brain
+ - 2015-06-19 23.34.11.ndpi
/git/ome-files-performance/scripts/run_benchmarking: line 63: -: command not found
[sbesson@ome-c6220-1-4 ~]$
```

I am merging this PR as it contains the minimal code for the new tiling benchmark, builds correctly and unlocks future work. Next incremental step will be to fix the execution of the code and the execution scripts semantics as suggested  in https://github.com/openmicroscopy/ome-files-performance/pull/44#discussion_r120314320 and captured in https://trello.com/c/v0jZwWJv/13-scripts-and-naming. Minor comments about the variable names can also be fixed in a follow-up PR.