Add compute instance creation for provisioning of managed vnet

[oloomi]

Description

This PR includes the following changes:

• Creates an AML compute instance (with no public IP) after the workspace is created, to trigger provisioning of managed vnet
• Upgrades API version from 2024-04-01-preview to 2024-04-01

Type of Change

• [ ] Non-module change (e.g. CI/CD, documentation, etc.)
• [ ] Azure Verified Module updates:
  • [ ] Bugfix containing backwards compatible bug fixes
  • [ ] Someone has opened a bug report issue, and I have included "Closes #{bug_report_issue_number}" in the PR description.
  • [ ] The bug was found by the module author, and no one has opened an issue to report it yet.
  • [x] Feature update backwards compatible feature updates.
  • [ ] Breaking changes.
  • [ ] Update to documentation

Checklist

• [x] I'm sure there are no other open Pull Requests for the same update/change
• [x] My corresponding pipelines / checks run clean and green without any errors or warnings
• [x] I did run all pre-commit checks

[vyta]

Should this be required for managed vnet in order to enable container builds on the compute instance? Or should the container builds be a separate config option?

[oloomi]

Should this be required for managed vnet in order to enable container builds on the compute instance? Or should the container builds be a separate config option?

That's right, when the Azure Container Registry for our workspace is behind a virtual network, it can't be used to directly build Docker images. Instead, we need to configure our AML workspace to use a compute cluster or compute instance to build images.

Let's use this compute instance as the image builder and it serves both purposes: provisions the managed network, and it can be used for image building.

We need to use an Azure CLI command (or a Python code) after the workspace and compute instance are created to update the workspace and set the image-build-compute parameter to the created compute instance.

Setting the imageBuildComput property in the workspace at this stage creates a cycle and causes Terraform to fail.