Feature request for the Evaluation module

[anlarro]

1. It would be nice to have the possibility to calculate the evaluation measures within a given mask. For example, I have a brain CT and the corresponding inferred synthetic-CT, I want to evaluate the performance by computing MAE, RMSE, etc. but only within the head mask thus excluding the background air.
2. I've also noticed that when running "evaluation" for a regression application, the results of MAE, RMSE, etc. are affected by the volume_padding_size and pixdim values assigned in the config file. I believe that this is an issue that should be resolved as the purpose of the evaluation is just to compare the inferred with the ground truth images.