Wrap cropAndResize with an op decorator

[dsmilkov]

Because we weren't wrapping cropAndResize using the op decorator, a user was seeing this stack trace when executing a model that had cropAndResize as an op:

engine.ts:197 Uncaught (in promise) TypeError: Cannot read property 'name' of null
    at e.runKernel (engine.ts:197)
    at Object.cropAndResize_ [as cropAndResize] (image_ops.ts:296)
    at executeOp$8 (image_executor.ts:65)
    at executeOp$16 (operation_executor.ts:65)
    at p (graph_executor.ts:323)
    at e.processStack (graph_executor.ts:309)
    at e.<anonymous> (graph_executor.ts:294)
    at callbacks.ts:17
    at Object.next (callbacks.ts:17)
    at i (callbacks.ts:17)

---

This change is 

[timekarma]

After taking https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@0.14.2/dist/tf.min.js
I am now getting this error when
loading the Model ok
then error is seen here
const pixels = tf.fromPixels(image);
const res1 = await model.executeAsync(pixels.reshape([1, ...pixels.shape]));

tensor_array.ts:116 Uncaught (in promise) Error: TensorArray : Could not write to TensorArray index 0,
because the value dtype is int32, but TensorArray dtype is float32.
at e.write (tensor_array.ts:116)
at tensor_array.ts:162
at Array.forEach ()
at e.writeMany (tensor_array.ts:162)
at e.scatter (tensor_array.ts:252)
at control_executor.ts:127
at callbacks.ts:17
at Object.next (callbacks.ts:17)
at callbacks.ts:17
at new Promise ()

[timekarma]

The above error is coming in 0.14.2 release.
In 0.14.1 it works fine. I tested it using the standard object detection example

[nsthorat]

@timekarma would you mind filing an issue here: /~https://github.com/tensorflow/tfjs/issues/new

I think this PR is unrelated.

Also, if you could host your model for us to take a look at we can fix it ASAP!