RSConnect: speed up pin_search by allowing using guid behind the scenes

Currently, python pins with RSConnect needs to look up the user and content guid for actions like

fs.info("michael.chow/mtcars")

however, looking up the user guid requires iterating through all users. This is an expensive operation that R pins does as little as possible by using the guids whenver possible (through a local cache file).

Let's support something like

# similar protocol format to sqlalchemy uris
# may need to use "rsc://guid+<user_guid>
fs.info("rsc+guid://<user_guid>")                  # GET v1/users/{guid}
fs.info("rsc+guid://<user_guid>/<content_name>")   # GET v1/content?name=<content_name>?user_guid=<user_guid>

fs.info("rsc+guid://content/<content_guid>")       # GET v1/content/<content_guid>

Note that we must meet these criteria...

• Files are cached correctly (e.g. could use the guids for everything in the cache)
• Users do not need to specify guid in pin_* methods
• We could allow users to specify guid in pin_* methods if it's not a lot of work. Otherwise, could start by having pins use it as an internal fast path.

Start by tracing pin_search and the RSConnect user and content GUID lookup paths, then inspect how files are cached. Use the proposed rsc+guid URI forms and the listed API requests as the behavioral guide. Done means cached files can use GUIDs, ordinary pin_* calls still work without user-supplied GUIDs, and the internal fast path is covered.