Funny, you said you'd ask two questions and then asked three. I agree with the above statements that a good medical school and residency program will give you thorough training on how to evaluate research and determine whether it is reliable. Giving a detailed answer to the question "how do you do it?" would take years and could be the subject of an entire master's degree - the short answer is, "that's part of our years of training."
1. When you need primary research, you search PubMed and find an article that has been thoroughly peer-reviewed, ideally in a respected journal (which would imply that several impartial experts held it to high standards of rigor). When you need the latest clinical guidelines, you go to UpToDate or a similar resource. When you want general principles, you look at a reliable textbook (my favorite is Medscape/Emedicine, which isn't a textbook per se, but might as well be). If that's not enough to answer a clinical question, I might read the APA guidelines or talk to a more senior physician whose opinion I respect.
2. Certain sources have reputations for being unbiased. UpToDate and Medscape are such examples, since they are written by experts, cite detailed sources, and don't have financial conflicts of interest. Drug company literature, to the contrary, is obviously biased and should not be trusted blindly without evaluating other unbiased sources. Journal articles should be evaluated for bias by looking at the reputation and/or impact factor of the journal (higher-impact journals will have a more rigorous peer review process to ensure scientific quality), the financial conflicts of interests of the authors (which are required to be reported with the journal), and the methods of the study itself to ensure that they accounted for potential confounding factors (ignoring certain key confounders is an easy way for a drug company to manipulate the literature).
3. Some journals (only the minority) come with a numeric rating of articles. Feedback from experts should be incorporated into the journal after the review process and shortcomings of the research are generally addressed in the discussion section; some newer online journals (i.e. F1000, PeerJ) include a section for comments. In general, if you're looking for a vague "rating" of an article, that usually equates to the impact factor of the journal, since any given journal will generally have certain quality standards that will be consistent across articles in that journal - for instance, if an article is in American Journal of Psychiatry or Biological Psychiatry (both have impact factor >10), you can assume that the quality of that article was around a "10." If an article is in a more subspecialized journal with lower impact, you can guess that it wasn't good enough to get into the more general higher-impact journal. That doesn't mean that there aren't great studies in low-impact journals - the higher-impact journals may have just thought that the article wasn't interesting enough, even if it was good... for instance, if an article is in Brain Stimulation (a respected journal for neuromodulation subspecialists with impact factor around 4.5), I might assume that it's a good article, but somebody didn't think that it was significant enough to warrant getting into Am J Psych - in other words, it is not too interesting to a general psychiatrist, but may be interesting to a neuromodulation expert.
That's a brief overview. In the real world, the process is much more complicated and is based on a lot of prior experience with evaluating literature... for instance, I know that a small study in a low-impact journal is less likely to be reliable than a large study in a high-impact journal, since I've seen that the latter study is more likely to be replicated and its findings are more likely to be successful when I actually try to apply them to patient care.
