TLDR; There’s no restrictions to the amount of traffic, but certain bandwidth restrictions apply.
Traffic is “shaped”, meaning that any one connection’s use of bandwidth over a certain limit is scheduled with a lower priority than other traffic — but only the traffic going over the threshold is rescheduled.
In reality, this limits the bandwidth available to any one connection, should it go over the threshold, only if other traffic eats up sufficient bandwidth. Packets landing in the back of the queue doesn’t implicitly mean lowering the bandwidth for that connection. It more likely increases latency, and caps the TCP window scaling because of it. Effectively, you may end up limited in bandwidth as a side-effect of fairly balanced scheduling.