Add Model Target dataset commands

[adamtheturtle]

Summary

The CLI covers the VWS target management, Query and VuMark APIs, but has no commands for the Model Target Web API. That API has an asynchronous dataset-generation workflow which is well suited to a CLI: create a standard or advanced dataset, poll its status, download the generated zip, and delete the dataset.

This is blocked on client support in vws-python (VWS-Python/vws-python#3119). vws-python-mock already implements the workflow — OAuth2 client credentials, standard and advanced dataset creation with validation errors, status, download and delete — so the commands can be tested end to end without bespoke test doubles.

Suggested scope

• Add commands for creating standard and advanced Model Target datasets.
• Add a command for retrieving dataset status, and a wait/poll option or command mirroring the existing wait_for_target_processed style.
• Add commands for downloading the generated dataset zip to a path and for deleting a dataset.
• Add options for the OAuth2 client ID and secret, and for the base URL, following the existing option and environment-variable conventions.
• Map Model Target API errors to useful exit codes and messages, as _error_handling.py does for VWS errors.
• Add documentation, shell-completion coverage and a news fragment.

Acceptance criteria

• The full create → poll → download → delete lifecycle is available from the CLI for both standard and advanced datasets.
• Tests exercise the commands against vws-python-mock, including authentication and validation failures.
• Documentation and --help output describe the new commands and options.