Allow queuing or setting an "appetite level" in Autopilot mode

### Feature summary

I need better ways to control how long Autopilot should keep going without stopping

### What problem are you trying to solve?

Autopilot often stops after finishing a "slice" when I want it continue until it's fully done, and I need ways to keep it going. When in interactive mode, I often queue the same prompt over and over. Things as simple as "continue" or a long prompt that repeats the goals to avoid drift after 25 iterations of the same thing. The thing is that Autopilot doesn't let you queue - it's _supposed_ to just continue until it's done, but I run long complex plans where even Autopilot needs a lot of cheering on to finish executing the plan, despite my instructions not to stop until it's fully executed.

Here's a sample where it stops after finishing a "slice" (it always looks like that when it happens):

The quickest hack would be to queue a bunch of "continue" like 5 times to keep it going overnight, and cross fingers I wake up to a completed plan. However, since there's already a plan which seems to be internally divided in "slices", I think we could probably add an "appetite level" instead. This is starting to sound like a pizza metaphor, but the idea would be to have controls to set how much of an appetite you have for going all the way or stopping after some time. If you're on a limited budget, you probably don't want it to keep going for 24 hours, but if you've got a solid plan and really want it to keep doing the work, you could set it to be very "hungry".

I'm not sure what would be the best way to set this appetite. Maybe a number of slices, a number of hours, or just a generic "appetite level" with "low/medium/high/ultra" similar to the thinking levels, but used for how far you want to go in autopilot mode. This might be a lot better than my hack of queuing tons of "continue" prompts in interactive mode.

### Proposed solution

_No response_

### Workflow impact

_No response_

### Installation context

_No response_

### Additional context

_No response_