Use different schema for pgstatsmon-specific functions and types

[kellymclaughlin]

This came up in the CR discussion for #14. Part of that change introduces a custom return type for a new function to collect vacuum progress information. Postgres does not currently support a CREATE TYPE IF NOT EXISTS statement so we must instead use DROP TYPE IF EXISTS ... CASCADE before the CREATE TYPE statement in pgstatsmon. This should be ok, but there is a risk that if some other part of the monitored system introduced an identically named type or function as one used by pgstatsmon then we could experience unexpected behavior or failures in either pgstatsmon or the monitored system. To avoid or at least limit this risk I suggest we explore moving the pgstatsmon-specific functions and types into a pgstatsmon schema namespace. It would require the queries using the functions to directly address the namespace rather than relying on the default public namespace, but I believe that should be a pretty minor change.