update base model and add as_dict support for base model

[tadelesh]

resolve #1990

model base change and test is here: Azure/azure-sdk-for-python#31028

also, fix #2034, #2035, #2036

[tadelesh]

I don't think we should replace it with as_dict bc:

1. body could be non model
2. body could be a MutableMapping[str, Any] that mix simple dict and Model
3. json.dumps(Model.as_dict()) will go through all models' attributes twice, it is not effiecient

[tadelesh]

cc @iscai-msft @lmazuel @johanste

[tadelesh]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[iscai-msft]

we exclude_none True right now, it's just that it's currently handled in the JSON encoder. I feel it'll be clearer if we just pass in True here

[tadelesh]

I'm a little confused. Let me give an example that you could help to correct:

class TestModel(Model):
    prop1: str
    prop2: str
    prop3: str
model = TestModel({"prop1": None, "prop2": NULL})
json.dumps(model, cls=AzureJSONEncoder, exclude_none=True)  # the result should be {"porp2": null}
json.dumps(model, cls=AzureJSONEncoder, exclude_none=False)  # test result should be {"prop1": null, "prop2": null}

[iscai-msft]

what is this added code for? I don't know if we should access a private model in the typing namespace

[tadelesh]

It is for type alias like:

union MyNamedUnion {
  one: Model1,
  two: Model2,
}