Add Model Serving Endpoints to databricks-bundles package

[yarri-bryn-cla]

Changes

This PR adds support for Model Serving Endpoints in the Python bundles package, allowing users to define and manage serving endpoints using Python code.

• Adds ModelServingEndpoint resource type
• Updates codegen to handle enums with dashes
• Add cross-namespace re-exports in codegen so that several modules align under the model_serving_endpoint namespace (e.g. EndpointTag)
• Updates Resources for model serving endpoints

Why

Model Serving Endpoints can be defined in conventional DABs but not in pyDABs. Managing Mosaic AI Gateway endpoints via yaml-only configurations becomes verbose quite quickly. Python support enables configuration in a user-friendly manner across workspace environments (e.g. via OmegaConf)

The cross-namespace re-export enables use of EndpointTag and other methods in a manner compatible with the existing api. The openAPI schema defines some attributes of model serving in an adjacent namespace (model_serving_endpoints and serving).

Enum fix ensure that provider inputs like amazon-bedrock resolve correctly in python.

Tests

• Full test suite passes (172 tests, cov 92%)
• New TestCase in /python/databricks_tests/core/test_resources.py for model_serving_endpoint passes
• All Makefile commands succeed in addition to test suite
• Codegen + Build produces a usable wheel artifact tested in a DAB project, successfully deploys to Azure Databricks.

[github-actions]

An authorized user can trigger integration tests manually by following the instructions below:

Trigger:
go/deco-tests-run/cli

Inputs:

• PR number: 4340
• Commit SHA: 82d965dc40259ba69c007962c8f5e9c3f9654641

Checks will be approved automatically on success.