Improve vector search documentation

[guimachiavelli]

Given its strategic importance, we need to improve vector search documentation by:

• break the page into at least three different documents:
  • a tutorial helping users activate vector search
  • guides walking users through setting up the different embedders
  • an exhaustive reference for the required settings/parameters (could be integrated into the relevant reference pages)
• create new explanation articles:
  • what is AI-powered search?
  • keyword vs semantic vs hybrid search
• update terminology (e.g. prefer AI-powered search over vector search)

Vector search documentation is almost completely contained within the experimental feature page. This has become problematic as the page is starting to become bloated. It tries to do too many different things, but does not do any of them as well as it should. Additionally, its terminology is starting to drift from the way marketing is communicating about vector search (e.g. AI-powered search seems to be the preferred term now).