Add option to use the original image entrypoint when creating session containers

Sometimes users may want to run the original image entrypoint in addition to our `entrypoint.sh`.It is "easy" to just execute yet another command from `entrypoint.sh` by passing the original entrypoint to the container via configuration files or environment variables, but there are several technical considerations, though:

- The image's original entrypoint command/script may assume the "root" privilege, while our `entrypoint.sh` switches to the plain `work` user.
- We should keep our kernel runner to become ready even if the original entrypoint script takes a long time or fails, to make the container "usable" at least.\* The execution of the original entrypoint script should be done asynchronously and we should have a way to expose its execution log.

JIRA Issue: BA-121