[ML] Do not retry model deployment. (#185012)

When deploying a trained model, we should not attempt retries as the
deployment may take longer than the es client's default 30s timeout.

x-pack/plugins/ml/server/routes/trained_models.ts

@@ -579,10 +579,15 @@ export function trainedModelsRoutes(
       routeGuard.fullLicenseAPIGuard(async ({ mlClient, request, response }) => {
         try {
           const { modelId } = request.params;
-          const body = await mlClient.startTrainedModelDeployment({
-            model_id: modelId,
-            ...(request.query ? request.query : {}),
-          });
+          const body = await mlClient.startTrainedModelDeployment(
+            {
+              model_id: modelId,
+              ...(request.query ? request.query : {}),
+            },
+            {
+              maxRetries: 0,
+            }
+          );
           return response.ok({
             body,
           });