Error handling

[TNanukem]

The platform should delete algorithms that generate an error in order to keep the file system clean. Also, there are improvements to be made on how to show the researcher that an error happened.

[TNanukem]

The algorithm still not deleted because by some reason, when the error algorithm achieves the feature phase, even tough the correct algorithm took the semaphore and is running, the correct algorithm throws an error because of the name of the archive that ends up with an extra .java in it.

I'm not sure why this error happen, since this .java only appears when the error algorithm has it's auxList made before the semaphore is taken. Maybe blocking the entire for loop, not only the simulation would solve it, however, this seems overly complicated to do.