Workcells should only be running 1 single tasks at a time

[aaronchongth]

Semantically, workcells should only be operating on 1 task at a time, and our current approach for parallel workcell tasks, will cause the workcell orchestrators to handle multiple tasks at the same time, while requiring the plumbing that is waiting for signals (whether it is transporter or AMR arriving)

Some ideas and constraints were discussed,

• workcells should only ever be allowed to run 1 task
• the signalling and waiting for signal should be refactored into the orchestrators themselves, instead of being part of the BT. This will provides a few benefits,
  • users will no longer be required to know about all the signalling required to get a task going
  • workcell tasks will be isolated and self-contained, which makes behaviors and debugging more reproducible and repeatable
• for AMR related tasks, we can consider setting up the RMF task dispatch directly from the workcell BT, so that each workcell task is responsible for its own input and output. This goes along with Refactor Workcell Bidding + Assignments into respective BT nodes #66