1 add Setter for the Model in PhysicsModels:

SetXSModel
SetEquivlalenceModel
SetIrradiationModel

Many MLP modification have been made it may not be 100% working for the moment... But the modification should improve the MLP model performance... and it by far...

This prevent to recreate a new inputTTree at each call of the MLP model, it will load the reader when the model is loaded, and link it the some class parameter. 
At each call of the MLP model, it only update the values of the internal parameter (linked to the reader) and launch the TMVA execution... which should be much faster than recreating a new inputTTrre, a new reader and new internal parameter link to it at each call of the MLP model....



git-svn-id: svn+ssh://svn.in2p3.fr/class@797 0e7d625b-0364-4367-a6be-d5be4a48d228