docs: Update get_started notebook and RAG page (WIP)

[wukaixingxp]

What does this PR do?

Update get_started notebook to replace old apis with new apis and similarly for the RAG page

Test Plan

[reluctantfuturist]

Can we update the existing getting started guide instead? I think we're better off having one canonical notebook

[franciscojavierarceo]

so i started exploring redoing this in the spirit of the existing quickstart, which aims to show value with basic agentic rag in as few lines of possible.

NOTE: As a brief aside, we used to be able to embed the demo script in the markdown with sphinx but it appears that is no longer true. I'm sure there's a way we could embed the script and compile it or something but that's a todo for another time.

Here's what the OpenAI Responses version of our quickstart looks like:

import io, requests
from openai import OpenAI

url="https://www.paulgraham.com/greatwork.html"
client = OpenAI()

tmp_vector_store = client.vector_stores.create(name="temp_store")

response = requests.get(url)
pseudo_file = io.BytesIO(str(response.content).encode('utf-8'))
file_id = client.files.create(file=(url, pseudo_file, "text/html"), purpose="assistants").id

client.vector_stores.files.create(vector_store_id=tmp_vector_store.id, file_id=file_id)

resp = client.responses.create(
    model="gpt-4o-mini",
    input="How do you do great work?",
    tools=[{"type": "file_search", "vector_store_ids": [tmp_vector_store.id]}]
)

print(resp)

It's quite short and to the point and also highlights citations—I think it shows a lot of value in a little bit of code.

IMO we should highlight this quickstart with Responses but also show it with Chat Completions using the Vector Store Search API with traditional RAG (even though that's not agentic and is more code).

It would then follow, in my opinion, that the Quickstart would lead to the Detailed Tutorial, which could then lead to the RAG Deep Dive that would outline more sophisticated examples (Probably ones that expose some of the other parameters within the VectorStores.Files.create and VectorStores.Search APIs.

But that's just a thought on how this could flow together to be maybe fluid.