Wrong character set detection logic for Oracle DB

Logic for checking character encoding with Oracle DB is wrong in server and frontend.

Oracle has two character set variables that we check: NLS_CHARACTERSET & NLS_NCHAR_CHARACTERSET.

We check that both of these are UTF8, which is wrong.

1. Oracle uses UTF8 name for CESU-8. Proper UTF8 encoding is called AL32UTF8.
2. We assume that both NLS_CHARACTERSET & NLS_NCHAR_CHARACTERSET have the same value. This may not be the case.

Suggest allowing AL32UTF8 & UTF8 (CESU-8). The latter should be supported for backwards compatibility with existing installations.