Add manual generic AI agent test

[ocots]

Summary

• Rename the caller workflow to AIHello.yml.
• Expose the AI provider, model, and prompt as workflow_dispatch inputs.
• Call the generic ai-agent.yml reusable workflow.
• Use the generic hello agent from Handbook/ai-agents/.
• Keep the initial test configured for the Albert provider and its deepseek model alias.

Related PRs

• Generic agent definition in Handbook: Add first generic AI agent definition Handbook#20
• Generic reusable workflow in CTActions: Add generic AI agent workflow CTActions#72

This PR is the end-to-end entry point. After the three PRs are merged, the branch references can be changed from the development branches to main.

Test plan

• Validate the workflow file as YAML.
• Run git diff --check.
• Run AI Hello from the GitHub Actions UI with the configured organization secrets.
• Confirm that the workflow retrieves the generic Handbook agent and receives a response from the selected provider.

Generated with Devin

[ocots]

This PR is the end-to-end test entry point:

• It calls the reusable workflow from CTActions PR #72.
• That workflow retrieves the agent defined in Handbook PR #20.
• It can be launched manually from the GitHub Actions UI with the configured organization secrets.

The workflow intentionally references the development branches until the three PRs have been reviewed and merged.

[ocots]

The caller is now provider- and agent-independent:

• It invokes CTActions PR #72 through the generic ai-agent.yml workflow.
• That workflow retrieves the generic hello agent from Handbook PR #20.
• The provider remains configurable, with Albert selected for this first end-to-end test.

A future Claude test can use the same caller shape once a Claude adapter is added to the shared CTActions workflow.