Splitting wait and initialization logic to allow for docker image extensions

Summary

I ran into a problem when trying to build a test version of the Gitlab image. I wanted to run some commands after the Gitlab instance was configured, before the Gitlab tail and wait was called. Unfortunately, I couldn't do this without completely copying the wrapper file. To get around this, I split up the init logic from the tail/wait logic. This way users trying to extend the docker image can override the cmd call and call their own logic while maintaining the normal initializing logic simply by calling this init file.

I've attempted to test this locally, but I'm missing some directories (RELEASE). Any suggestions to resolve this and fully test this would be helpful!