[Fix #3390] Process is not completed when node is retriggered (but end node is executed)

[fjtirado]

Calling getNodeInstance over containernode creates another instance, hence the previous node container instance is there when the engine checks if the process can be completed (so it does not complete it)
Therefore, as an alternative, a existing instance with that node id is searched, if found, that is the parent container instance where to recreate the node instance to be retriggered. If not found, we call getNodeInstance over the container node.

[elguardian]

@fjtirado this is incorrectly merged You need two commiters approval before merge, not much of a hurry you have. That is the procedure. You also reused an issue.
@porcelli @martinweiler @baldimir

[fjtirado]

@elguardian I reused the original issue because it was not completely working, as can be easily seen following the code changes, the explanation in the description and the additional IT test check. I do not really see any problem, other than I forgot to reopen it (to be honest I do not know if that is possible, which I know is that all commits related to a problem inherited from V7 are in the same github issue, which is good for inquiries)
I believe two approvals was not a strict requirement and one approval was enough, but from your reaction I guess thats not longer the case. My apologies, next time, I will wait to have two approvals (in this particular case, I was not in a hurry, the reason I merged is because I really believe one commiter was enough and the PR was green)
To avoid that happening again, I recommend enforcing this two commiter restriction using github rules.

[caponetto]

LGTM! (I can't approve a merged PR though)