feat: add Vector Search 2.0 + ADK travel agent notebook

[kazunori279]

Description

Adds embeddings/vector-search-2-travel-agent.ipynb, an end-to-end walkthrough of building a Gen AI travel agent on Agent Platform Vector Search 2.0 and ADK: ingest real Airbnb listings into a Collection with auto-embeddings, run semantic / text / hybrid search with metadata filters, then wrap hybrid search as a tool an ADK agent calls on its own.

The notebook comes from google/adk-samples, where it was deleted in google/adk-samples@63bfb33c along with the rest of the deprecated notebooks. embeddings/ is a better home for it, next to vector-search-2-intro.ipynb.

Changes made while moving it:

• Rebranded to Agent Platform and repointed the doc links at docs.cloud.google.com, matching vector-search-2-intro.ipynb.
• Added the repo notebook header: license cell, badge table, author, Overview.
• %pip instead of !pip, and tqdm is now installed rather than only imported.
• Listings with no price are dropped instead of defaulted to 0.0. About a third of the Inside Airbnb rows have no price, so the old fillna(0.0) made every "under GBP X" filter match them, and the agent then reported listings it had no price for.
• find_rentals(filter=...) renamed to metadata_filter so it stops shadowing the builtin; the agent instruction follows.
• The run_debug calls bind their result, so the readable trace is no longer followed by a screenful of the returned event list.

embeddings/README.md also picks up three pre-existing "Vertex AI Vector Search" -> "Agent Platform Vector Search" rewrites, because the forbidden-pattern check scans the whole changed file.

Testing

Executed end to end against google-adk 2.8.0 and google-cloud-vectorsearch 0.11.2 (400 listings, real project): ingestion, all three search types, the three agent queries and the cleanup cell run with no errors, and the agent produces the expected compound filters, e.g. {"$and": [{"neighborhood": {"$eq": "Hackney"}}, {"price": {"$lt": 200}}, {"instant_bookable": {"$eq": "t"}}]}. Outputs are stripped from the committed notebook.

Checklist

• Follow the CONTRIBUTING Guide.
• Ensure the tests and linter pass (./scripts/format.sh).
• You are listed as the author in your notebook or README file.
• Your account is listed in CODEOWNERS for the file(s).

---

🤖 Generated with Claude Code

[holtskinner]

Thanks for adding! I just found out that Vector Search 2.0 is now called "Agent Retrieval" https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/vector-search-2/overview

Could you update this notebook to use the new name?