2) Should I be at the same distance while shooting to get same noise grain size? (to get same framing, I move forward using the 5D) or do I need to move back some more?
Ultimately, it doesn't matter - or it matters a lot. The reality is that there's not a good way to effectively compare the noise. The closest you're going to get is to do what you've done (move the camera to compensate for the crop factor and achieve the same framing), then compare both images without resizing them. Yes, the T2i image will 'look bigger'. The other option is the more standardized way that testers like DxO use, which is to print everything at a fixed size.
This issue gets to the heart of the argument over whether more MP are inherently bad. The converse to what you're finding is that people who moved from a 40D to a 7D (or 60D) think the new camera is noisier, so the smaller pixels must be bad. But the reality is, if you downsample the 18 MP image to the 10 MP of the 40D, there' not really different. The only reason smaller pixels are inherently bad is the effect of diffraction.
You're also running the test with two very different sensor generations. I've done a similar comparison of the 5DII to the 7D (similar in age, and the 7D is the same sensor as the T2i), and I find that when you don't change framing, but rather crop the 5DII image to that of the 7D AoV, the 5DII has slightly less noise, and the 7D is slightly sharper. Mainly a wash. But if I change the distance (or zoom a lens) so I'm capturing the same framing with the full sensor in both cases, the 5DII wins for both noise and sharpness.