There’s no capping for Tunnels that we have set on purpose anywhere. In fact, I use many Tunnels on a daily basis and download tens MiB/sec from some of them.
If you experience those problems again, can you re-run the tunnel with --protocol http2 and let us know if it made any difference?
Quick Tunnels are testing quic protocol, which we are invested in improving, but we need this kind of feedback when things go wrong (and the problem becomes obviously rooted in http2 vs quic).