fix: debounce output clearing during cell re-execution

[nstrayer]

When a notebook cell is re-run, the runtime immediately clears outputs before sending new ones. This caused the output container to briefly disappear and reappear, producing a distracting flash.

lotsa-bouncing.mov

To address we defer the clear by a small amount when execution is active so fast-completing cells never flash. We also flush the deferred clear immediately when execution ends.

Result:

Screen.Recording.2026-04-10.at.1.40.28.PM.mov

Release Notes

New Features

• N/A

Bug Fixes

• Fixed notebook cell outputs flashing briefly when re-running a cell

QA Notes

@:notebooks @:positron-notebooks

1. Open a Python notebook
2. Run a cell that produces output (e.g. print("hello"))
3. Re-run the same cell -- output should update smoothly without a flash/flicker
4. Try with a slow cell (e.g. import time; time.sleep(2); print("done")) -- output should clear only after the delay, not flash
5. Clear outputs manually (right-click > Clear Outputs) -- should clear immediately without delay

[github-actions]

E2E Tests 🚀
This PR will run tests tagged with: @:critical @:notebooks @:positron-notebooks

^{readme  valid tags}

[seeM]

Yay thank you! These sorts of visual flickers really bother me - glad I'm not alone.

1. Should this rather happen in the UI layer? There are other consumers of cell outputs and they probably don't want this debounced behavior. E.g. could create a derived/debounced observable from outputs in the NotebookCodeCell. Maybe the existing debouncedObservable works?

2. Out of scope of this change, but can we also debounce the execution status indicator and the execution count change?

   

[nstrayer]

Good call. The approach I took (a new useDebouncedObservedValue() hook) made it trivial to fix the status indicator and count as well.

[seeM]

Nice, super clean and works well! I also debounced the execution order in 796221d, figured it was faster to commit directly than comment