feat: Druid JDBC driver.

[gianm]

This patch adds a Druid JDBC driver that uses the regular endpoint, /druid/v2/sql/. It is recommended over the Avatica driver for a variety of benefits, including: driver-side connection management that doesn't require sticky load balancing, streaming results, support for SET statements, and support for non-native SqlEngines.

In addition to the new module, there are core changes:

1. TIMESTAMP parameters to parameterized queries, when provided as strings
   with zone offsets, are now correctly interpreted in the provided zone.

2. SQL now correctly treats function names in JDBC {fn ...} escapes
   case-insensitively.

3. SQL now correctly handles the JDBC escape {fn now()}.

[FrankChen021]

Severity	Findings
P0	0
P1	0
P2	2
P3	0
Total	2

Reviewed 76 of 76 changed files.

---

This is an automated review by Codex GPT-5.6-Luna(max)

[FrankChen021]

[P2] Closing during HTTP startup does not cancel the server query

Statement.close() can race with executeSql() before httpClient.runQuery() returns. With no result set yet, closeCurrentResultSet() only clears currentSqlQueryId; when the request returns, executeSql() closes the new result stream but never calls cancelQuery. The server-side query can remain running after the statement is closed. Preserve and cancel the in-flight query ID, and add a blocked-request test.

[FrankChen021]

[P2] Network timeout stops at response headers

The request uses BodyHandlers.ofInputStream(), so HttpClient.send() can return after headers arrive while QueryResultsIteratorImpl reads the body later. If the server sends headers and then stalls, ResultSet.next() or hasNext() can block indefinitely despite a finite Connection.setNetworkTimeout(). Enforce a body-read or idle deadline, or cancel the request on expiry, and add a stalled-body test.