Combine the docker files for training and for deploying

[Gautam-Rajeev]

Combining the docker files for training (offline training):

Currently we have two docker compose files for training the model - one for bootstrapping and another for training

We need to able to combine these into one file such that we only need to run one docker compose. The training should happen only after the bootstrap container has completed and the container exits (with exit code 0 )

Combining the docker files for deploying the model once trained:

Currently we have two docker compose files for deploying the model - one for uploading the model and current state into redis - upload and another for running the model itself such that it can give recommendations - inference

We need to able to combine these into one file such that we only need to run one docker compose. The upload should complete, however this container does not exit as redis is kept open and then the inference should be run.
Currently, we need to run upload in a terminal instance and then the inference in a separate terminal instance.