[Ralph Orchestrator] Manager for continuous ralph systems with separate worktrees (Rebase hat)

My intention is to build a manager for spawning and running ralph systems on a continuous basis, much of what I want I intend to build as a extension of either your ralph crate or an alternative. My ideal workflow with ralph is as follows: 1. I upload large user-tasks to a central branch (`main` or `develop`) on my repo's remote. 2. The manager continuously spawn new ralphs for each user-task on the server. 3. Each ralph works on an issue in its own worktree 4. Once it's completed, rather than spawning a merge-ralph task, it pushes its branch to the remote (ideally along with a summary document of the changes for me to review). 5. I review the changes and then either merge or reject them into the remotes `main` branch (or some other custom central branch like `develop`) 6. The ralph server sees the update to this central branch and then rebases all pending branches to the central branch. - This could be maybe be achieved by spawning some special rebase ralph, or maybe a the e