Make mlops module independent of train and serving modules

Previously, MLOps orchestration was possible within a Lamber. Now with new dependencies, they are too big to fit in the lambda. The native dependencies and PyTorch explode the size of the total deployment signficantly

Start by tracing the dependencies between the MLOps, train, and serving modules, then reproduce the Lambda packaging size problem described in the issue. Identify which native dependencies and PyTorch are pulled into MLOps deployments. Done means MLOps orchestration can be packaged independently without those unnecessary dependencies.

Written by the indexing model from the issue text.