feat: add support for updating existing Agent Engine instances

[hanorik]

This PR introduces the ability to update an existing Vertex AI Reasoning Engine (Agent Engine) instance using the adkgo deployment tool, removing the need to always deploy a new instance when changes are made.

Key Changes

• CLI Flags: Added two new persistent flags to the agentEngine deployment command:

  • --update: A boolean flag to trigger the update workflow.
  • --instance_name: Specifies the full resource name of the Agent Engine instance to update (required if --update is set).

• Update Workflow: Implemented gcloudUpdateAgentEngine() to handle the update process:

  • Automatically extracts the GCP region from the provided --instance_name to set up the correct regional endpoint.
  • Packs and uploads the updated source code archive.
  • Retrieves and registers available class methods via ListClassMethods to ensure the instance specification matches the code.
  • Utilizes a FieldMask (spec.source_code_spec, spec.class_methods) to selectively update the instance via the aiplatform client.

• Deployment Logic: Updated deployOnagentEngine to route to the update logic when the flags are present.

[gemini-code-assist]

Code Review

This pull request introduces the capability to update an existing Agent Engine instance by adding --update and --instance_name flags to the deployment command. It implements the gcloudUpdateAgentEngine method, which utilizes the ReasoningEngineClient to update the source code and class methods of a specified instance. A review comment suggests that the update request should also include DeploymentSpec and AgentFramework fields to ensure consistency with the creation logic and to correctly manage environment variables and framework identifiers.