d385b8f597
The Bazel build part of tensorflow builds can really use as many cores as is available - to avoid blocking a small machine for literally hours while it churns away, mark the build as big-parallel so we can schedule it to run quickly on a machine with lots of cores.