test(router): cover provider over-budget filtering path

tests/test_litellm/router_strategy/test_budget_limiter_hotpath.py

@@ -182,6 +182,45 @@ async def test_async_filter_deployments_keeps_provider_without_max_budget(
     assert filtered_deployments == healthy_deployments
 
 
+@pytest.mark.asyncio
+async def test_async_filter_deployments_filters_provider_over_max_budget(
+    disable_budget_sync, monkeypatch
+):
+    provider_budget = RouterBudgetLimiting(
+        dual_cache=DualCache(),
+        provider_budget_config={
+            "openai": BudgetConfig(budget_duration="1d", max_budget=100.0),
+        },
+    )
+
+    await provider_budget.dual_cache.async_set_cache(
+        key="provider_spend:openai:1d", value=150.0
+    )
+
+    healthy_deployments = [
+        {
+            "model_name": "gpt-4o-mini",
+            "litellm_params": {"model": "openai/gpt-4o-mini"},
+            "model_info": {"id": "deployment-1"},
+        },
+    ]
+
+    monkeypatch.setattr(
+        provider_budget,
+        "_get_llm_provider_for_deployment",
+        lambda deployment: "openai",
+    )
+
+    with pytest.raises(ValueError, match="Exceeded budget for provider openai"):
+        await provider_budget.async_filter_deployments(
+            model="gpt-4o-mini",
+            healthy_deployments=healthy_deployments,
+            messages=[],
+            request_kwargs={},
+            parent_otel_span=None,
+        )
+
+
 @pytest.mark.asyncio
 async def test_async_filter_deployments_does_not_recompute_provider_when_resolved_none(
     disable_budget_sync, monkeypatch