feat: shared iteration budget across parent + subagents

Subagent tool calls now count toward the same session-wide iteration
limit as the parent agent. Previously, each subagent had its own
independent counter, so a parent with max_iterations=60 could spawn
3 subagents each doing 50 calls = 150 total tool calls unmetered.

Changes:
- IterationBudget: thread-safe shared counter (run_agent.py)
  - consume(): try to use one iteration, returns False if exhausted
  - refund(): give back one iteration (for execute_code turns)
  - Thread-safe via Lock (subagents run in ThreadPoolExecutor)
- Parent creates the budget, children inherit it via delegate_tool.py
- execute_code turns are refunded (don't count against budget)
- Default raised from 60 → 90 to account for shared consumption
- Per-child cap (50) still applies as a safety valve

The per-child max_iterations (default 50) remains as a per-child
ceiling, but the shared budget is the hard session-wide limit.
A child stops at whichever comes first.

tools/delegate_tool.py

@@ -194,6 +194,10 @@ def _run_single_child(
         # Build progress callback to relay tool calls to parent display
         child_progress_cb = _build_child_progress_callback(task_index, parent_agent, task_count)
 
+        # Share the parent's iteration budget so subagent tool calls
+        # count toward the session-wide limit.
+        shared_budget = getattr(parent_agent, "iteration_budget", None)
+
         child = AIAgent(
             base_url=parent_agent.base_url,
             api_key=parent_api_key,
@@ -215,6 +219,7 @@ def _run_single_child(
             providers_order=parent_agent.providers_order,
             provider_sort=parent_agent.provider_sort,
             tool_progress_callback=child_progress_cb,
+            iteration_budget=shared_budget,
         )
 
         # Set delegation depth so children can't spawn grandchildren