test: remove unstable llamaindex args evaluator

Author: radu-mocanu

packages/uipath-llamaindex/template/evaluations/eval-sets/evaluation-set-default.json

@@ -5,7 +5,6 @@
   "evaluatorRefs": [
     "evaluator-llm-judge-output",
     "evaluator-tool-call-order",
-    "evaluator-tool-call-arguments",
     "evaluator-tool-call-count"
   ],
   "evaluations": [
@@ -20,7 +19,6 @@
           "expectedOutput": "{\n  \"response\": \"Yes, it's quite good weather for a walk in Paris! The current conditions are:**Temperature**: 18°C (64°F) - mild and pleasant spring weather **Wind**: 12 km/h (7 mph) - light, manageable breeze **Conditions**: Partly cloudy - perfect for a walk without being too sunny or too gray\"\n}"
         },
         "evaluator-tool-call-order": null,
-        "evaluator-tool-call-arguments": null,
         "evaluator-tool-call-count": null
       },
       "createdAt": "2026-04-09T10:00:00.000Z",

packages/uipath-llamaindex/testcases/template-agent/src/assert.py

@@ -67,13 +67,6 @@
 assert score == 1.0, f"Tool call order score should be 1.0, got: {score}"
 print("  ✓ tool call order: perfect score (1.0)")
 
-# Tool call arguments evaluator should have perfect score (1.0)
-assert "evaluator-tool-call-arguments" in run_results_by_evaluator, \
-    "Missing evaluator result for: evaluator-tool-call-arguments"
-score = run_results_by_evaluator["evaluator-tool-call-arguments"]["result"]["score"]
-assert score == 1.0, f"Tool call arguments score should be 1.0, got: {score}"
-print("  ✓ tool call arguments: perfect score (1.0)")
-
 # Tool call count evaluator should have perfect score (1.0)
 assert "evaluator-tool-call-count" in run_results_by_evaluator, \
     "Missing evaluator result for: evaluator-tool-call-count"