Expose ModelCheckpoint parameters in convenience configuration functions

[jdeschamps]

Problem

If users want to set the parameters of the checkpoointing, they need to modify the values directly into the Pydantic configuration:

from careamics.config import create_care_configuration

config = create_care_configuration(...)
config.training_config.checkpoint_callback.every_n_epochs = 5

We could instead allow a checkpoint_kwargs or better yet the ModelCheckpoint Pydantic model in the API for the convenience functions.

[jdeschamps]

We should be more explicit when passing it to CAREamist:

careamics/src/careamics/careamist.py

Line 235 in cf3a094

"training configuration (see TrainingConfig)."