[8.5][ML] Increase timeout for _infer in PyTorchModelIT (#91875)

This is a manual backport of #91853 as the code is a bit
different in the 8.5 branch.

Author: dimitris-athanasiou

x-pack/plugin/ml/qa/native-multi-node-tests/src/javaRestTest/java/org/elasticsearch/xpack/ml/integration/PyTorchModelIT.java

@@ -989,15 +989,15 @@ private Response infer(String input, String modelId, TimeValue timeout) throws I
     }
 
     private Response infer(String input, String modelId) throws IOException {
-        Request request = new Request("POST", "/_ml/trained_models/" + modelId + "/_infer");
+        Request request = new Request("POST", "/_ml/trained_models/" + modelId + "/_infer?timeout=30s");
         request.setJsonEntity("""
             {  "docs": [{"input":"%s"}] }
             """.formatted(input));
         return client().performRequest(request);
     }
 
     private Response infer(String input, String modelId, String resultsField) throws IOException {
-        Request request = new Request("POST", "/_ml/trained_models/" + modelId + "/_infer");
+        Request request = new Request("POST", "/_ml/trained_models/" + modelId + "/_infer?timeout=30s");
         request.setJsonEntity("""
             {
               "docs": [ { "input": "%s" } ],