Container run steps should allow leaving shell and command blank to to use the image's entrypoint
I
Isolated Meerkat
For container Run step, use the image entrypoint defined in the dockerfile instead of requiring shell and command fields. This works only for background step at the moment. It would be useful for Run steps as well. The use case is to run some tests where the image is already set up to run them. It should be blocking (not background) since it is for running smoke tests.
Log In
I
Isolated Meerkat
Also setting the initial working directory to "/harness" should be optional since the image entrypoint may expect a particular working directory.