Most existing Ops rely on DefaultChunker when they wish to multithread a process. This process is more complicated to introduce in scijava-ops because:

1. It requires all ops that want to multithread to require an ExecutorService as a parameter, something that could/should be injectable by the OpService.
2. The tests for this Op introduce junitbenchmark; it would be nice if we did not have to introduce this.

There are a few different approaches that we can take:

1. Create some sort of static method that mimics the behavior of Chunk
2. Add support for what is going on in Unify multithreading handling imglib/imglib2-algorithm#81

Would it be possible to create a special mutlithreading OpDependency that, upon discovery, the matcher could inject its ThreadService into? i.e. if you had a

@OpDependency(name = "thread.chunker")
private NullaryComputer<Chunk> chunkerOp;

Ops could find a Computer that takes an ExecutorService as input and wrap that into a NullaryComputer by injecting its ExecutorService?