Switch maxNumTokensWithReasoningModel to run on o4-mini and drop gpt-3.5

Author: hjh320

tests/tazureChat.m

@@ -141,10 +141,7 @@ function maxNumTokensWithReasoningModel(testCase)
             % them after the model deployed is not a guarantee), that is a
             % somewhat painful distinction.
             testCase.verifyWarningFree(@() generate( ...
-                azureChat(DeploymentID="gpt-35-turbo-16k-0613"), ...
-                "What is object oriented design?", MaxNumTokens=23));
-            testCase.verifyWarningFree(@() generate( ...
-                azureChat(DeploymentID="o1-mini"), ...
+                azureChat(DeploymentID="o4-mini"), ...
                 "What is object oriented design?", MaxNumTokens=23));
         end