[Experimental] Screenshot-based browsing

[ryanhoangt]

End-user friendly description of the problem this fixes or functionality that this introduces

• Include this change in the Release Notes. If checked, you must provide an end-user friendly description for your change below

---

Give a summary of what the PR does, explaining any non-trivial design decisions

This PR is to:

• Add a screen-shot based browsing mechanism based on Anthropic's computer use feature.

---

Link of any specific issues this addresses

Fix #4570

[ryanhoangt]

Asking the agent to draw a square, and it currently looks pretty weird...

[mamoodi]

@ryanhoangt just a gentle ping to check if this is still a work in progress

[ryanhoangt]

Yes, I'll try to get back to this after finishing the other PR.

[github-actions]

This PR is stale because it has been open for 30 days with no activity. Remove stale label or comment or this will be closed in 7 days.

[mamoodi]

@ryanhoangt this was just tagged as stale. The related PR seems merged. Will you be continuing this when time permits now?

[ryanhoangt]

Yeah I discussed with the team and we should close this PR for now. Given some limitations of Browsergym to implement computer use and Aditya has added visual browsing capability to CodeAct in #6464, and we can probably try that first before moving on to computer use, which will require some changes to the runtime to use Playwright directly instead of via Browsergym.