Support base_transform_job_name param in [Estimator|Model].transformer()

[athewsey]

Describe the feature you'd like

When creating a Transformer from a Model via Model.transformer() or Estimator.transformer(), standard Transformer constructor arguments should be supported and in particular - base_transform_job_name.

How would this feature be used? Please describe.

When creating a transform job from an Estimator or Model, users want to be able to control the prefix of the job name without having to explicitly set the full job_name: Similar to how Estimator supports base_job_name for training jobs to which a timestamp is automatically appended.

Today, the transformer() functions on Model and Estimator don't seem to support passing through base_transform_job_name.

I have found that the following seems to work, but it relies on undocumented / non-guaranteed behaviour that setting the property on the transformer is sufficient for it to be used:

transformer = model.transformer(...)
transformer.base_transform_job_name = "my-cool-job-base"
transformer.transform(...)

Describe alternatives you've considered

Possible options could include:

1. Supporting open-ended **kwargs in Estimator & Model's .transformer() functions, to allow pass-through of all relevant properties
2. Explicitly adding the base_transform_job_name parameter (and perhaps taking a survey of any other missing params.

Additional context

I found this while using xgboost in algorithm mode, so am using plain Estimator. My workflow goes via Model before creating the transformer, with estimator.model().