Open source wall scheduling

Currently there is no sharable scheduler for wall training/inference.

In tandem with the shared scheduler we should also add a mechanism for basic general execution using OpenCue/Deadline.

This would open up wall training to the public.

No files, tests, or entry points are identified. Begin by clarifying the shared scheduler and basic OpenCue/Deadline execution requirements, including how wall training and inference become publicly sharable. Done means the project has an agreed design and implementation path for both scheduling and general execution.

Written by the indexing model from the issue text.