Make the executor service of ParallelFlow configurable

[fmbenhassine]

As of v0.1, the ParallelFlow implementation uses an internal executor service to run work units in parallel. The idea behind this design is to be able to create and shutdown the executor service internally and free the user from having to do that.

However, the drawback of this design is that every parallel flow uses its own executor service, which is obviously not efficient since the creation of thread pools is expensive. Moreover, using an internal executor service is not open for customizations.

This is a trade-off between:

1. Lifecycle management of the executor service (creation/shutdown)
2. Efficiency (re-using the same executor service for all parallel flows) and customization of the executor service

I believe trading option 1 for option 2 is a wise decision. In the end, the executor service lifecyle management is a couple of lines of code, for example:

ExecutorService executorService = Executors.newFixedThreadPool(10);
// pass executorService to parallel flow
executorService.shutdown();