ZTF_ATAT: early stopping object configured to track loss_validation/mix but mode set to "max"

[HexaPulsar]

Issue regarding ATAT_ALeRCE training at branch main:

training.py: early stopping object configured as track='loss_validation/mix' but mode ='max' . For consistency with tracked variable it mode should be set mode= 'min' or track should change to a variable that increases over time.

EarlyStopping(
            monitor="loss_validation/mix",
            min_delta=0.00,
            patience=args_general["patience"],
            verbose=False,
            mode="max",
        )

should be

EarlyStopping(
            monitor="loss_validation/mix",
            min_delta=0.00,
            patience=args_general["patience"],
            verbose=False,
            mode="min",
        )