Add support for registered models for direct deployment

[shreyas-goenka]

Changes

This PR adds support for registered models, matching the same behaviour as TF. This PR also adds test server handles for registered models and UC catalogs, as was necessary for the integration test.

Why

To complete coverage for direct deployment.

Tests

New integration test.

[eng-dev-ecosystem-bot]

Run: 18162343450

	Env	✅​pass	🔄​flaky	🙈​skip
✅​	aws linux	320		537
✅​	aws windows	321		536
✅​	aws-ucws linux	436		433
✅​	aws-ucws windows	437		432
✅​	azure linux	320		536
✅​	azure windows	321		535
✅​	azure-ucws linux	436		432
🔄​	azure-ucws windows	435	2	431
✅​	gcp linux	319		538
✅​	gcp windows	320		537

Test Name	azure-ucws windows
TestFilerWorkspaceNotebook	🔄​flaky
TestFilerWorkspaceNotebook/rJupyterNb.ipynb	🔄​flaky

[denik]

As discussed offline, would be great to use acceptance/bundle/resources/<resource_name>/ schema.

bundle/deploy is really focussed on deploy command but we're also interested in plan/json plan/destroy/summary.

another aspect is to use smaller tests which is good for perf and debugging:

See how this test is organized:

 ~/work/cli/acceptance/bundle/resources % tree -d volumes
volumes
├── change-comment
├── change-name
├── change-schema-name
├── remote-change-name
├── remote-delete
└── set-storage-location

[shreyas-goenka]

I'll address this in a followup PR.

[denik]

Appreciate making it both local and integration test, that adds to confidence.

[denik]

recording "bundle debug plan" would be nice as well.

[shreyas-goenka]

I'll address this in a followup PR.

[denik]

recording non-GET requests would be useful to ensure terraform and direct implementations match. If there is too much difference in payload, it can either be cleaned up or we can just record paths without the payload.

[shreyas-goenka]

I'll address this in a followup PR.

[denik]

The implementation looks solid, there are comments on test organization, can be done here or as a follow up.