Separate new project + classification from the analysis

[avouros]

• The user should be able to load all his datasets (training and probe trials) if the experimental procedures (arena and platform size and location and trial timeout) are the same.

• The software should classify all the trajectories and export the raw classification result. Probably here some analytics can be performed: e.g. count / average thigmotaxis, incursion, etc per trial, per day, per animal, etc.

• The Friedman test will be performed at the last stage with an appropriate interface for selecting on which configure and animals it should be performed on (with appropriate checking if it can be performed; the groups needs to have equal animals which performed equal number of trials..