fix(ci): force --table output + post-LOAD probes in mysqlx-soak setup

The Bring up infrastructure step pipes the SQL block to `mysql` via
docker exec -i with stdin redirected from a heredoc. With stdin not
attached to a tty `mysql` switches to a silent default that suppresses
header + row output for non-error queries, so the existing SELECT
COUNT(*) probes left no trace in the CI log — making it impossible to
tell whether the lazy-refresh nudge ran, whether LOAD MYSQLX TO RUNTIME
returned errors, or whether the install_*_from_admin path saw zero
source rows.

Two surgical changes:

* `mysql --table` forces boxed output regardless of stdin shape, so
  SELECT results land in the runner log.
* Add a final UNION SELECT that counts rows in
  runtime_mysqlx_users / runtime_mysqlx_routes / runtime_mysqlx_
  backend_endpoints after the LOAD statements run. Zero in any of
  those columns localises the bind failure to install_*_from_admin
  having seen empty source rows.

test/tap/groups/mysqlx-soak/setup-infras.bash

@@ -59,7 +59,10 @@ fi
 # what the X-protocol client authenticates against; the X-specific
 # overrides (allowed_auth_methods, default_route, backend_auth_mode)
 # live in mysqlx_users.
-docker exec -i "${PROXY_CONTAINER}" mysql -u${ADMIN_USER} -p${ADMIN_PASS} -h127.0.0.1 -P6032 <<SQL
+# --table forces boxed output even on stdin so the SELECT COUNT(*) and
+# any error from LOAD MYSQLX ... TO RUNTIME end up in the CI log.
+# Without it mysql's stdin-driven default suppresses non-error output.
+docker exec -i "${PROXY_CONTAINER}" mysql --table -u${ADMIN_USER} -p${ADMIN_PASS} -h127.0.0.1 -P6032 <<SQL
 DELETE FROM mysql_users WHERE username='${TEST_USER}';
 INSERT INTO mysql_users (username, password, active, default_hostgroup, frontend, backend)
     VALUES ('${TEST_USER}', '${TEST_PASS}', 1, 10, 1, 1);
@@ -96,6 +99,14 @@ SELECT COUNT(*) FROM runtime_mysql_servers;
 LOAD MYSQLX USERS TO RUNTIME;
 LOAD MYSQLX BACKEND ENDPOINTS TO RUNTIME;
 LOAD MYSQLX ROUTES TO RUNTIME;
+
+-- Post-LOAD probes: surface what the plugin actually installed so a
+-- failure to bind the listener can be diagnosed from the CI log
+-- (counts of zero here mean install_*_from_admin saw empty source
+-- rows despite the SELECT-COUNT(*) nudge above).
+SELECT 'runtime_mysqlx_users' AS table_name, COUNT(*) AS rows FROM runtime_mysqlx_users
+UNION ALL SELECT 'runtime_mysqlx_routes', COUNT(*) FROM runtime_mysqlx_routes
+UNION ALL SELECT 'runtime_mysqlx_backend_endpoints', COUNT(*) FROM runtime_mysqlx_backend_endpoints;
 SQL
 
 # Wait for the listener to bind.