Inconsistent termination for infeasible MINLPs in parallel QG

[ppalkar]

If the root NLP is infeasible, the parallel QG (mcqg) algorithm throws an error (not QG) due to a slightly different logical flow than QG. This has to be fixed by capturing the infeasibility status of the root NLP and terminating the algorithm at appropriate function in ParQGHandler and McQG.