Support more JDBC-aware databases in JDO TCK

The TCK does not support databases except for Derby, even though the RI does.

The primary blocker is that the TCK uses an embedded Derby database.

The proposed solution is to use a JDBC connection to the database instead of an embedded Derby database.

There are a few issues:

• Before running the TCK, the database will need to be started, which may involve a manual operation or a database-dependent script. And after the TCK test completes, the database will need to be shut down.

• The database schema need to be created, which currently uses a Derby-specific program. This program needs to be replaced by a program that reads a SQL command file containing the SQL commands to create the schema. And the schema needs to be created manually or automatically based on metadata.