Docker script refactoring and rewrite to Python

[vladak]

This change rafactors the scripts used for Docker. The initial motivation was to avoid the non-existent-configuration exception thrown by the web app during the initial startup of a container based on the opengrok image.

Some notable changes:

• no special casing between initial and non-initial indexing
• the sequence goes like this:
  1. deploy
  2. bare config is created (to avoid above mentioned exception)
  3. indexer is started in the background (waits for the Tomcat to come up before it enters the infinite loop)
  4. Tomcat is started on foreground
• to preserve the locking I added filelock to the indexer program

This is a step towards #3221.

Eventually the start.sh script can be rewritten to Python and use native interfaces to opengrok-sync, opengrok-deploy.

I used the opportunity to overhaul everything - the main Docker program is now based on the opengrok-sync and rewritten into Python.

[vladak]

I used the opportunity and changed start.sh to use opengrok-sync and then rewritten it to Python. The way it works now is that the sync runs the mirroring/indexing/... in parallel for each projects and once this is all done, it will print the logs for each project to the console.

[vladak]

Incidentally, #3421 describes a use case when project's repositories set is changed. This change does not support such use case even though it can be easily added.

[vladak]

thanks @ahornace !