I worked out that if you wanted to be able to resolve features on the planet surface at the 1000 km scale (some fraction of a continent size), at a distance of 1 million light years, and you want to do this at a wavelength of around 10 microns, which is roughly where the blackbody spectrum for something with an effective temperature of 300 K peaks, then you would need a telescope with a circular aperture of diameter approximately 115 MILLION kilometres. That's almost the distance between Earth and the sun. So you'd need telescope equal in size to half the Earth's orbit. And that STILL doesn't guarantee that you'd be able to see any emission from that planet (because it might be too faint, or it might be swamped by foreground sources or detector noise). If I worked things out right (which I may not have), such a planet would have a bolometric flux of ~2e-28 W/m2, leaving about 10 microwatts of power incident on the telescope. I'll leave you guys to convert that into a magnitude or into janskys or whatever it is you use.