add ability to resume training from checkpoint

[Youssef2430]

if the training crashed for whatever reason, add ability to resume training from a checkpoint selected

[Youssef2430]

you can now resume training your model from the last valid checkpoint by running this command

python src/main.py -r -train -v -checkpoint 27000