This is quite tricky.
The ‘obvious’ solution is an emonPi in the house, measuring the PV array and diverted energy, and an emonTx in the garage measuring the grid connection and the car charger. Unless there are many or thick walls in the way, or a strong interfering signal, the 433 MHz link should work over 30 - 40 m. This would get around the poor WiFi problem.
I assume you’re using Robin’s own pcb, and not his software on an Arduino/emonTx with an add-on triac switch. Do you have the radio module (RFM12B or RFM69CW) fitted?
The ‘burst’ nature of Robin’s diverter means that the “discrete sample” sketch, the default in both the emonTx and emonPi, won’t read the diverted power particularly accurately, and in turn that’s going to cause inaccuracies all along the line. EmonLibCM will solve that problem, but while it’s on test now in the emonTx, no attempt has yet been made to integrate it into the emonPi. In the short term, the inaccuracy will be significant, but over time, it should even out.
If you have the radio module, then it will send power and the total diverted energy. That will get around the inaccuracy problem at the house end, but it will still be there (but partly swamped by the more constant loads) in the grid connection.
You could run a cable to the garage. Only yesterday, we got a report of a Canadian friend using CAT 5 cable over a longer distance than yours. You could then have an emonTx (2 cable inputs from the garage, PV and immersion from the house) and emonBase all in the house.
In the long term, I’d hope to be able to say that both emonPi and emonTx will have the continuous monitoring software that will remove the problem. It will certainly be available, and hopefully very soon, for the emonTx, but there might be problems with the emonPi.
The temperature monitoring should not pose a problem, both emonTx and emonPi will accept those sensors.