PostgreSQL: better documentation for statement caching

[jaskij]

As is, the only explanation for how deadpool_postgres handles caching prepared statements I'm aware of is the discussion in #75 . Between that and looking at the source, I understand the behavior and it's fine, but:

• it's hard to find
• it's not a contract in the way documentation is, and relying on undocumented behavior and a four year old issue is bad practice

Better documentation - especially the information about ClientWrapper handling the caching internally - would making getting started with deadpool_postgress that much easier.

[bikeshedder]

A PR to add this documentation would be highly appreciated! :heart: