Run a long running cell or run the entire notebook and then start run by line.
We need to ensure we handle these cases.

Since run by line will send code directly to the kernel and bypass the cell execution queue.
E.g. if you have 10 cells and run all, then for the last cell we hit RBL, then the last cell could start debugging immediately after the first cell has completed.

I think we need to add a method executeHidden into the KernelExecution class so that it uses the queue & debugging starts after all queued cells have been completed.
Should be an easy enough fix, else we end up in a very weird spot.