update minimum required capacities for GPT models in checkquota.sh; remove premature satisfaction guard logging in human_approval_manager.py

Author: Ayaz-Microsoft

infra/scripts/checkquota.sh

@@ -24,8 +24,8 @@ echo "✅ Azure subscription set successfully."
 
 # Define models and their minimum required capacities
 declare -A MIN_CAPACITY=(
-    ["OpenAI.GlobalStandard.gpt-5.4-mini"]="${GPT_5_4_MINI_MIN_CAPACITY:-0}"
-    ["OpenAI.GlobalStandard.gpt-5.4"]="${GPT_5_4_MIN_CAPACITY}"
+    ["OpenAI.GlobalStandard.gpt-5.4-mini"]="${GPT_5_4_MINI_MIN_CAPACITY:-100}"
+    ["OpenAI.GlobalStandard.gpt-5.4"]="${GPT_5_4_MIN_CAPACITY:-150}"
 )
 
 VALID_REGION=""

src/backend/v4/orchestration/human_approval_manager.py

@@ -275,14 +275,6 @@ async def create_progress_ledger(self, magentic_context: MagenticContext):
                     guard_round_budget,
                     uncalled,
                 )
-            if uncalled and within_budget:
-                logger.warning(
-                    "Premature satisfaction guard exhausted its round budget (%d) with "
-                    "agent(s) still marked uncalled: %s. Allowing termination to avoid an "
-                    "infinite loop (likely an author_name/participant-name mismatch).",
-                    guard_round_budget,
-                    uncalled,
-                )
             if uncalled and within_budget:
                 next_agent = uncalled[0]
                 logger.info(