Fix medium/low severity issues: script splitter, EXPLAIN, keyring, imports

execute_script() semicolon splitter (all drivers)
- Add BaseDriver._split_statements() that tracks single/double-quoted
  strings and -- / /* */ comments so semicolons inside procedure bodies
  are not treated as statement boundaries.
- Replace the naive sql.split(';') in all four drivers with this helper.
- MSSQL execute_script() also pre-splits on GO (case-insensitive, own
  line) so scripts pasted from SSMS work correctly.

MSSQL EXPLAIN
- SET SHOWPLAN_TEXT ON/execute/SET SHOWPLAN_TEXT OFF must be separate
  execute() calls; pyodbc rejects multiple statements in one call.
  Hold self._lock for the entire sequence to keep session state atomic.

Keyring (connections.py)
- Log warnings (with traceback) on load/save failures instead of
  silently returning "".
- save_password() now calls delete_password() when password is empty,
  so clearing a saved password actually removes the old keyring entry
  rather than leaving a stale one behind.

SQLite get_tables() O(N) lock acquisitions
- Run all COUNT(*) queries inside the same `with self._cur() as c:`
  block, reducing N+1 lock acquisitions to 1.

Unused import: remove psycopg2.extras from postgres_driver.py.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

app/drivers/mssql_driver.py

@@ -233,21 +233,23 @@ class MSSQLDriver(BaseDriver):
             return [], [], c.rowcount
 
     def execute_script(self, sql: str) -> list:
+        import re
         results = []
-        stmts = [s.strip() for s in sql.split(';') if s.strip()]
-        for stmt in stmts:
-            try:
-                with self._cur() as c:
-                    c.execute(stmt)
-                    if c.description:
-                        cols = [d[0] for d in c.description]
-                        rows = [tuple(r) for r in c.fetchall()]
-                        results.append((cols, rows, len(rows), ""))
-                    else:
-                        results.append(([], [], c.rowcount,
-                                        f"{c.rowcount} row(s) affected"))
-            except Exception as e:
-                results.append(([], [], 0, f"Error: {e}"))
+        batches = re.split(r'^\s*GO\s*$', sql, flags=re.IGNORECASE | re.MULTILINE)
+        for batch in batches:
+            for stmt in self._split_statements(batch):
+                try:
+                    with self._cur() as c:
+                        c.execute(stmt)
+                        if c.description:
+                            cols = [d[0] for d in c.description]
+                            rows = [tuple(r) for r in c.fetchall()]
+                            results.append((cols, rows, len(rows), ""))
+                        else:
+                            results.append(([], [], c.rowcount,
+                                            f"{c.rowcount} row(s) affected"))
+                except Exception as e:
+                    results.append(([], [], 0, f"Error: {e}"))
         return results
 
     # ── Table data CRUD ───────────────────────────────────────────────────────
@@ -305,9 +307,20 @@ class MSSQLDriver(BaseDriver):
     # ── Server tools ──────────────────────────────────────────────────────────
 
     def explain_query(self, sql: str) -> tuple:
-        with self._cur() as c:
-            c.execute(f"SET SHOWPLAN_TEXT ON; {sql}; SET SHOWPLAN_TEXT OFF")
-            return ["Plan"], c.fetchall()
+        with self._lock:
+            cur = self._connection.cursor()
+            try:
+                cur.execute("SET SHOWPLAN_TEXT ON")
+                cur.execute(sql)
+                cols = [d[0] for d in cur.description] if cur.description else ["Plan"]
+                rows = cur.fetchall()
+                cur.execute("SET SHOWPLAN_TEXT OFF")
+            finally:
+                try:
+                    cur.close()
+                except Exception:
+                    pass
+        return cols, rows
 
     def get_process_list(self) -> tuple:
         with self._cur() as c: