I’ve been reading that plugs that return power usage (I have the third reality ones and peanut plugs with updated firmware) and that they can really clog up a zibgee network with a lot of messages. Is this true and at what point would it be a problem?
I have about 35 lights and 20 plugs, just upgraded my controller to a UZG-01 and went to Z2M from ZHA.
Always useful to cite sources, IMHO.
There are several ways to judge traffic from simple to complex, and probably with corresponding level of accuracy.
On the simple end, if you are using Mosquitto as your MQTT broker, it shows you the number of MQTT message sent and received. Most other MQTT brokers have similar instruments.
With these numbers, you can write a simple bash script to see how often each type of your power plugs is publishing a MQTT message. I have no experience with the Peanut Plug, however I have used a number of others, Zigbee, Zwave, Wifi and have not seen a device that samples faster than single digits per second and publishes even less, often about 1 message per second.
As I said, you can get perhaps more accurate numbers by looking down into Zigbee2MQTT and further by looking down into Zigbee stack/packets.
You have a number of plugs, however, it is good to remember that Zigbee runs on a 2.4 GHz based carrier (fairly good road). Zigbee mesh can be complex, however, I think you would be hard pressed to be pushing your envelop.
A good back of the envelop ( and significant other acceptance) measure, is to Zigbee bind a switch to a light on your network and ‘press the button’ and watch the light 
Good hunting!
