Allow arbitrary strings in model version

### Describe your suggestion

I'm trying to implement semantic versioning rules ("MAJOR.MINOR.PATCH") for all Azure Machine Learning assets but I'm struggling since the version of a registered model must be a positive integer. This is unlike data assets or environments, where the version can be an arbitrary string. Is there any reason for this restriction in models or is it something that could be fixed in the future?

### Additional details

_No response_

The issue names no file, test, or entry point. Start by investigating Azure Machine Learning model registration and versioning, then compare the behavior with data assets and environments. Done means determining whether arbitrary semantic-version strings are supported and identifying the affected API or example changes.

Written by the indexing model from the issue text.