High level comment, taking this LRO as an example: https://github.com/googleapis/python-aiplatform/blob/dev/google/cloud/aiplatform/models.py#L211

When we receive an LRO from the GAPIC API call, it's already initialized and polling so we may not want to initialize a separate LRO and have two LROs polling for the same result. Unless we want to delete the initial LRO which I think adds too much complexity for what we want.

Instead, I think we want a Class that takes the LRO initialized by the GAPIC library as an attribute, exposes methods that are relevant to the MB usage (result, add_done_callback).

Additionally, we have a requirement to be able to update a MB object with resource properties once the LRO is complete. In the example linked above that would look something like:

lro = api_client.upload_model(...)
mb_lro = LRO(lro)
empty_model = cls.init_from_lro(api_client, mb_lro)
return empty_model

The init_from_lro on Model would look something like:

@classmethod
def init_from_lro(cls, api_client, lro):
  self = cls.__new__(cls)
  self.api_client = api_client
  self.lro = lro
  self.lro.add_update_resource_callback(
     self, lro_result_key='model, api_get=lambda name: api_client.get_model(name=name) )
  return self

So the LRO class would look something like:

class LRO:

  def __init__(self, gapic_lro: operation.Operation):
    self._lro = gapic_lro

 def add_update_resource_callback(self,
   resource_noun_obj: base.AiPlatformResourceNoun,
   result_key: str, api_get: Callable):
   def callback(result):
     result_value = getattr(result, result_key)
     resource = api_get(result_value)
     resource_noun_obj._gca_resource = resource
   self._lro.add_done_callback(callback)

Please make sure the kokoro build passes before merging.