Speedup hash computatation by calling pipdeptree just once

[azeller14]

Suppose we have a luisy Task TaskA that is located in packageA. Then for the hash creation of TaskA, we make a call to pipdeptree via a subprocess to console for each dependency listed in the requirements of packageA. This can be rather slow. Actually one call of pipdeptree is enough, we can sort the information inside python.