Explore execution path with multi-thread

[mox692]

For a complex program, loom often takes a long time to execute. Currently, loom explores the execution path in a single thread, but it would be cool if it had multi thread mode. Such features are also mentioned in #162 (comment).

The implementation is probably not trivial, but it would be worth considering.

EDIT: Well, since cargo's test runner already runs tests in parallel (thread per core) by default, perhaps we need to think of a clever parallelization method that would surpass this simple parallelization by cargo.