[Java][FlightRPC] Flight SQL JDBC driver improvements

Follow ups for ARROW-7744

- Rename internal classes to not imply everything is part of Flight RPC (e.g. ArrowFlightJdbcArray -> FieldVectorArray or similar)
- Don't throw bare exceptions (always provide some error context)
- Log a warning if the `arrow-flight:` URI scheme is used instead of `arrow-flight-sql:`
- Create a documentation page (that can be used for people approaching this from the JDBC side, not necessarily Arrow users)
- Document the connection string format and parameters
- Replace `// TODO` comments with `throw new UnsupportedOperationException()`
- Document how timestamp/time/date types are handled in converting between the two type schemas
- Document the type conversions in general
- [timestamp handling is suspect](https://github.com/apache/arrow/pull/13800#discussion_r938908230)
- Upgrade to JUnit5/AssertJ instead of JUnit4/Hamcrest
- Get rid of FreePortFinder
- Use a single (static?) RootAllocator with child allocators
- Remove all Guava usage (can we deny this with a lint?)
- See if `SqlTypes` can be folded into arrow-jdbc

**Reporter**: [David Li](https://issues.apache.org/jira/browse/ARROW-17729) / @lidavidm
#### Subtasks:
- [X] [[Docs][Java] Add documentation page for Flight SQL JDBC driver](https://github.com/apache/arrow/issues/32967)
- [X] [[Packaging] Add JDBC driver to release tasks](https://github.com/apache/arrow/issues/32976)

_{**Note**: *This issue was originally created as [ARROW-17729](https://issues.apache.org/jira/browse/ARROW-17729). Please see the [migration documentation](https://github.com/apache/arrow/issues/14542) for further details.*}

Beginne damit, die verbleibenden Folgeaufgaben von den beiden abgeschlossenen Teilaufgaben zu trennen, untersuche dann die im issue genannten Einstiegspunkte des Flight SQL JDBC driver und ermittle, welche noch nicht abgehakten Punkte weiterhin zutreffen. Erledigt bedeutet, dass die ausgewählten Verbesserungen, die Dokumentation, die Tests und die Aufräumarbeiten abgeschlossen sind, ohne dass das übergeordnete issue unklar bleibt.

Vom Indexierungsmodell aus dem Issue-Text verfasst.