Evaluator-optimizer — write a draft, let a judge critique it, revise — step 8 of 9

Build the full evaluator-optimizer. Write write_with_judge(task, max_iters) that:

• Maintains feedback = None (initial draft has no prior feedback).
• Loops up to max_iters times. Each iteration:
  • Calls gen(task, feedback=feedback) to produce a draft.
  • Calls judge(draft) to get a verdict dict.
  • If verdict["status"] == "PASS": return a dict {"draft": <draft>, "iters": <iteration count, 1-based>, "passed": True}.
  • Otherwise update feedback = verdict["feedback"] and continue.
• If max_iters is exhausted without a PASS, return {"draft": <last draft>, "iters": max_iters, "passed": False}.

The script will run two cases (one that PASSes on iter 2, one that exhausts the cap). Expected output:

passed=True iters=2 draft=Cats are small carnivorous mammals.
passed=False iters=2 draft=v2