tests: skip Issue161 test on Firebird 6

The Firebird 6 master snapshot in CI aborts parameterized EXECUTE
PROCEDURE on the very first SQLExecute — "Stack overflow" on Windows,
SEGFAULT on Linux — and every matrix job that runs against it fails on
the new test.  Locally the same test passes against the 6.0.0.1910
snapshot, and all FB 3 / 4 / 5 jobs pass green, so it isn't the Issue
#161 driver fix itself that regresses.  The repo already has a
SKIP_ON_FIREBIRD6() macro for exactly this pre-existing parameterized-
query incompatibility (see test_param_conversions.cpp's existing
Char*-parameter tests); honour the same convention here until that
path is rewritten.

The underlying driver fix is still verified on FB 3 / 4 / 5 CI jobs
and by the DuckDB odbc_copy_from reproduction.

Author: fdcastel

tests/test_param_conversions.cpp

@@ -286,6 +286,14 @@ TEST_F(ParamConversionsTest, NumericAsCharParam) {
 // execute, so multi-digit values collapsed to a single character and most rows
 // were silently lost (odbc-scanner issue #161).
 TEST_F(ParamConversionsTest, Issue161_SLongToVarcharViaStoredProcedure) {
+    // CI's Firebird-6 master snapshot aborts parameterized EXECUTE PROCEDURE
+    // with "Stack overflow" (Windows) or SEGFAULT (Linux) on the very first
+    // SQLExecute — the same parameterized-query regression already documented
+    // by SKIP_ON_FIREBIRD6().  The underlying driver fix is exercised on FB 3
+    // / 4 / 5 anyway, so punt this particular harness until the FB6
+    // parameterized-query path is rewritten.
+    SKIP_ON_FIREBIRD6();
+
     // Provision the sandbox: a VARCHAR-keyed table and an UPDATE-OR-INSERT SP.
     ExecIgnoreError("EXECUTE BLOCK AS BEGIN "
                     "IF (EXISTS(SELECT 1 FROM RDB$PROCEDURES WHERE RDB$PROCEDURE_NAME = 'ODBC_ISSUE161_SP')) THEN "