Oops. The "rate limit" I quoted is actually the limit on the DOCSIS modem here, not on the VPS. Probably not 'iptables' traffic shaping after all.
Using 'speedtest_cli.py' the max rate has been showing 100 Mbits/sec, but I discount that because the speedtest node appears to reside in the same data center as the VPS and is probably on the same LAN.
Nonetheless, the Tor bridge is showing a ridiculous low bandwidth value and it seems reasonable to figure out why.