When you have an a.c. adapter and a current transformer, each sample of current and voltage - and there are about 2500 pairs each second - are multiplied together to give the power at that instant, the power is then averaged. This is the real power, that you pay for.
If you average the current and voltage separately, as if reading them on two separate meters, you can only have what we call apparent power, which is the same or greater than the real power.
Your emonPi calculates both, but it only sends the real power IF you have an a.c. adapter. Otherwise it guesses the voltage and uses that to estimate the apparent power. This is done inside the “emon” part of the emonPi, long before the numbers arrive in emonCMS.
What Trystan was saying is, in emonCMS, you should divide the “power” you see by the estimate we use for voltage - 230 V - which will take you back a step to give you the current. You then multiply that current, still inside emonCMS, by the voltage you get from Carlo Gavazzi. That will give you a better value of apparent power, but you can never get real power by that method.
What form does the live voltage feed from the Carlo Gavazzi take? To get that voltage into emonCMS, you must send it via WiFi or Ethernet, in a form that emonCMS can understand. I’ve no concept of mqtt public, so I’ve no idea what it does.