docs for operations should link to their response types

Today we show something like this:

fn instance_list(&self) -> InstanceList<'_>
List instances

Sends a GET request to /v1/instances

Arguments:

limit: Maximum number of items returned by a single call
page_token: Token returned by previous call to retrieve the subsequent page
project: Name or ID of the project
sort_by
ⓘ
let response = client.instance_list()
   .limit(limit)
   .page_token(page_token)
   .project(project)
   .sort_by(sort_by)
   .send()
   .await;

Fine, but what does that return? To figure it out, I have to click InstanceList, then click send, then click its return type. It would be useful if we had that return type linked directly from the docs.

Some other nits here:

• could we make the example runnable / testable? (even if it's kind of janky)
• should paginated endpoints (such as this one) show the use of the stream() call rather than send()?
• should each of the arguments link to the appropriate type or to the function within the builder?

Start with the generated documentation for the instance_list operation and trace the links from InstanceList through send() to its response type. Define completion as linking the operation directly to its response type; the issue also raises separate questions about runnable examples, stream() for pagination, and links for arguments, which need scope decisions.