fix(academic-research): keep Agent Engine SDK versions in sync

[yaodong-shen]

Closes #1223.

Summary

The academic-research deployment script now pins the remote Agent Engine runtime to the locally installed google-cloud-aiplatform and google-adk versions. The Vertex AI requirement also includes the adk extra so the deployment receives the compatible framework integration.

This prevents the serialized AdkApp from being deployed with a different ADK API surface, which caused the canonical_on_model_error_callbacks type error reported in the issue.

The README now explains the version-parity requirement, and the deployment requirement builder has a credential-free unit test.

Tests

• .venv/bin/pytest -q tests/test_deployment_requirements.py — 1 passed
• .venv/bin/ruff check deployment/runtime_requirements.py deployment/deploy.py tests/test_deployment_requirements.py
• .venv/bin/ruff format --check deployment/runtime_requirements.py deployment/deploy.py tests/test_deployment_requirements.py
• .venv/bin/mypy deployment/runtime_requirements.py tests/test_deployment_requirements.py
• .venv/bin/codespell README.md deployment/deploy.py deployment/runtime_requirements.py tests/test_deployment_requirements.py
• .venv/bin/python -m compileall -q deployment/runtime_requirements.py deployment/deploy.py tests/test_deployment_requirements.py

With the checked-in lock file, the generated requirements are:

google-cloud-aiplatform[agent_engines,adk]==1.143.0
google-adk==1.27.4

A live Agent Engine deployment was not run because it requires a configured GCP project, storage bucket, credentials, and billable cloud resources.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.