currently there are 3 celery tasks that run inference, training and metrics. these are unbounded tasks which means queue is blocked if we reach max number of workers (currently set at 6 in docker-compose), that's 2 experiments.
add a small functionality to cancel running tasks for given experiment to be able to launch new experiments
track list of celery task ids for each experiment
feature pr: add additional way of running training and inference jobs that are infinite in its nature
feature pr: add additional way of running training and inference jobs that are infinite in its nature