Introduce AbstractSchemaManager::getCurrentSchemaName()

[morozov]

Currently, the logic of building qualified or unqualified table/view/sequence names during schema introspection is scattered across schema managers implementations an is inconsistent. On the one hand, this is okay since some database platforms support schemas, and others don't. However, at a high level, the introspection logic it the same for all platforms:

1. Query the introspection schema and group rows by the table to which they belong.
2. If the platform supports schemas, do not include the default schema into the name (so that accounts remains represented as accounts, not public.accounts on Postgres or dbo.accounts on SQL Server).

As part of the work on on introspecting table names as objects, I want to make this logic explicitly written in the code. And for that, the API of all schema managers should be consistent.