Answers

The importance of blind trials is to make sure that no factor varies systematically between the treatment group and the control group other than the treatment itself. If the patients know which group they are in, then this knowledge itself constitutes a systematic difference between the two groups. This might not seem like an important difference, but in fact simply believing that you are being treated can have a measurable effect for a wide range of medical conditions. Patients who are given a placebo to take often show a measurable improvement compared to patients who are given nothing at all. This is called the placebo effect.
Again, the importance here is to rule out any factors which might vary systematically between the treatment group and the control group. It has been shown that if the doctor knows whether the patient is in the treatment group or the control group, this knowledge can unconsciously alter their behaviour towards the patient and their perception of the patient's condition. In this case, if the doctor knows which group the patient is in, this might unconsciously influence their perception of slight changes in hair growth or of the severity of side effects. This makes the data provided by the doctor less reliable.
No, a controlled trial is unlikely to be possible in this case. We can't assign people randomly to two groups and make one group believe in superstitions. Hence the two groups (superstitious and non-superstitious) are self-selecting, and this introduces the possibility of bias. It may be that people with a predisposition to neurosis, depression or low IQ are more likely to end up in the superstitious group than in the non-superstitious group. This is why the correlation reported in the newspaper article does not establish a causal link.