DOI: A doctor who sucks at research.

My answer to both of these questions is that doctoring has a fairly low ceiling in terms of how much impact one can have. 

The way that healthcare systems in the developed world work means that "making a diagnosis" is a fairly small part of the day-to-day work of what doctors do. Once a certain level of experience is reached, diagnostic ability probably plateaus; some doctors might be better than others, but even if they are the impact of that discrepancy probably isn't that great. I've never (disclaimer- not been practicing very long) seen a "House MD" type situation where someone is desperately unwell because no one can make a diagnosis until some rogue genius comes along. Ready access to sophisticated investigations and interventions mean that clinical reasoning in that traditional style is probably less important than it was in the past. 

You're quite right that the average biomedical researcher probably doesn't have a massive social impact; the difference is that it is possible to be 10x or 100x (1000x?) better at research than a colleague, in a way that it's not really possible to a 10x better doctor than the average. 

This leads to a much more "long tail" distribution of impact in biomedical research, where a few researchers have massive impact, some have significant impact, and many have little impact. Presumably, when effective altruists suggest that research is more impactful than clinical practice, they mean that a) you might end up in that small group of very impactful researchers and b) given that possibility, and the impact if you did, the "expected value" of becoming a researcher is higher than that of becoming a doctor. 

If you were asking to compare impact between the median researcher (minimal impact) and the median doctor, the impact might be much closer.