Apparently no-one actually ran the mx_base_schedule, because the tests
in schedule itself were already failing. This updates it to be in line
with multi_mx_schedule again to make the tests pass again. Notably it
doesn't contain multi_mx_node_metadata and multi_extension. Because
those tests take long to run and the were not necessary to make
multi_mx_create_table pass again.
Test ALTER ROLE doesn't deadlock when coordinator added, or propagate from mx workers
Consolidate wait_until_metadata_sync & verify_metadata to multi_test_helpers