fix: improve stuck detection in UI mode

[enyst]

Allow the agent to continue after it gets stuck.

---

(Original openhands comment)

The issue has been resolved by making the stuck detection UI-mode aware:

1. In UI mode (with actual user):

   • Only consider history after the last user message
   • This allows users to naturally recover from stuck state by sending a new message
   • Agent needs to get stuck again (multiple steps) after user message to trigger error

2. In headless mode (automated/testing):

   • Keep existing behavior of looking at full history
   • Fail fast with error when stuck is detected

The implementation:

• Adds ui_mode parameter to stuck detection
• Uses headless_mode flag to determine default behavior
• Maintains backward compatibility
• Includes comprehensive tests for both modes

Fix: #5480

---

To run this PR locally, use the following command:

docker run -it --rm   -p 3000:3000   -v /var/run/docker.sock:/var/run/docker.sock   --add-host host.docker.internal:host-gateway   -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:1e739bd-nikolaik   --name openhands-app-1e739bd   docker.all-hands.dev/all-hands-ai/openhands:1e739bd