Update usage of deprecated API (get_weights) in TensorFlow 2 Optimizer.

[lalalapotter]

Encountered following error when I run LSTM model with orca pyspark tf2 estimator:

Traceback (most recent call last):
  File "lstm_pollution.py", line 33, in <module>
    steps_per_epoch= df.count() // batch_size)
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/bigdl/orca/learn/tf2/pyspark_estimator.py", line 199, in fit
    lambda iter: transform_func(iter, init_params, params)).collect()
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/pyspark/rdd.py", line 816, in collect
    sock_info = self.ctx._jvm.PythonRDD.collectAndServe(self._jrdd.rdd())
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/py4j/java_gateway.py", line 1257, in __call__
    answer, self.gateway_client, self.target_id, self.name)
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/pyspark/sql/utils.py", line 63, in deco
    return f(*a, **kw)
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/py4j/protocol.py", line 328, in get_return_value
    format(target_id, ".", name), value)
py4j.protocol.Py4JJavaError: An error occurred while calling z:org.apache.spark.api.python.PythonRDD.collectAndServe.
: org.apache.spark.SparkException: Job aborted due to stage failure: Could not recover from a failed barrier ResultStage. Most recent failure reason: Stage failed because barrier task ResultTask(12, 0) finished unsuccessfully.
org.apache.spark.api.python.PythonException: Traceback (most recent call last):
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/pyspark/python/lib/pyspark.zip/pyspark/worker.py", line 377, in main
    process()
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/pyspark/python/lib/pyspark.zip/pyspark/worker.py", line 372, in process
    serializer.dump_stream(func(split_index, iterator), outfile)
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/pyspark/rdd.py", line 2465, in func
    return f(iterator)
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/bigdl/orca/learn/tf2/pyspark_estimator.py", line 199, in <lambda>
    lambda iter: transform_func(iter, init_params, params)).collect()
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/bigdl/orca/learn/tf2/pyspark_estimator.py", line 196, in transform_func
    return SparkRunner(**init_param).step(**param)
  File "/home/cengguang/anaconda3/envs/bigdl2.1/lib/python3.7/site-packages/bigdl/orca/learn/tf2/spark_runner.py", line 349, in step
    "optimizer_weights": model.optimizer.get_weights()
AttributeError: 'Adam' object has no attribute 'get_weights'

I use tensorflow 2.11 and estimator.fit() API with argument model_dir specified. Seems that the program use deprecated API to get weights from optimizer. Please check if we need to update the usage.

[sgwhat]

Caused by a break change (in version 2.11.0) by Tensorflow team, still looking into a general solution.

[sgwhat]

keras-team/tf-keras#442 maybe a useful solution.