Update JNI binding layer comments to indicate that forward/execute is not thread-safe

Summary: Update doc comments on forward and execute to document that these functions are not thread-safe.

Differential Revision: D66144668

Author: GregoryComer

extension/android/src/main/java/org/pytorch/executorch/Module.java

@@ -66,17 +66,21 @@ public static Module load(final String modelPath) {
   /**
    * Runs the 'forward' method of this module with the specified arguments.
    *
-   * @param inputs arguments for the ExecuTorch module's 'forward' method. Note: if method 'forward'
-   *     requires inputs but no inputs are given, the function will not error out, but run 'forward'
-   *     with sample inputs.
+  * @param inputs arguments for the ExecuTorch module's 'forward' method. This function is
+   *     not thread safe and thus must not be called again while a previous inference is in
+   *     progress. Additionally, if method 'forward' requires inputs but no inputs are given,
+   *     the function will not error out, but run 'forward' with sample inputs.
+   *
    * @return return value from the 'forward' method.
    */
   public EValue[] forward(EValue... inputs) {
     return mNativePeer.forward(inputs);
   }
 
   /**
-   * Runs the specified method of this module with the specified arguments.
+   * Runs the specified method of this module with the specified arguments. This function is
+   *     not thread safe and thus must not be called again until while a previous inference is
+   *     in progress.
    *
    * @param methodName name of the ExecuTorch method to run.
    * @param inputs arguments that will be passed to ExecuTorch method.