Thank you for your help Alvaro - we really appreciate it.
The error in fact stopped this morning - we took downtime and ran a vacuum across all of our tables, and saw increased
autovacuum activity as well.
It looks like it bumped up the oldest multitxid to something other than 1 now:
postgres@site-db01a:~ $ /usr/pgsql-9.4/bin/pg_controldata /var/lib/pgsql/cmates/data | grep -i multi
Latest checkpoint's NextMultiXactId: 785051
Latest checkpoint's NextMultiOffset: 1446371
Latest checkpoint's oldestMultiXid: 575211
Latest checkpoint's oldestMulti's DB: 12998