I would guess that the MX appliances and the MR access points do a lot of their packet processing in CPU, due to the nature of what they do, and so limiting/shaping traffic is just part of the software. On the other-hand a lot of what switches do is performed in hardware/ASICs, which is a trade-off in flexibility for performance. I'm guessing that the ASICs in the switches can't do the limiting/shaping, and doing it in software on the CPU would likely cripple the switches. So I doubt you will ever see that capability in the traditional Meraki switches (in the MS390, who knows, that's a different beast).
(The throughput on even the most expensive MX is only 6Gbps, and on a the fastest MR is not going to exceed 10Gbps, whereas a small 24 port MS210 potentially needs to push 56Gbps (plus another 80Gbps for stacking) - hence why its done in ASICs, and thus limits the flexibility).