Pointing out a conflict of interest, by itself, is indeed an ad hominem. Whether someone has a conflict of interest or not has nothing to do with their argument being right or wrong. It only indicates that they might be willing to put forth their argument without exploring any supporting evidence (or willfully ignoring it). Besides, in this case, I personally think that is very unlikely.

Except that it does. Regardless of the ethos of a scientist, when a scientist sets out to research with a certain goal, "the positive effects of drinking coffee", the only case in which the result would be unfavourable to coffee would be when coffee has no positive effects whatsoever. This is as shown obviously not the case.

When you research "the effects of drinking coffee" it seems as though you are more neutral, but in fact there is still room for bias. Will you find every single effect of drinking coffee? How will you decide wether they are positive or negative? how will you decide the importance of these effects? Which effects will you research in more detail?

This weighing and ordering depends in part on the researcher, who will have a bias.

Ofcourse this is not a problem when there are lots of researchers all with their own biases, together they will paint a full picture. But if you only fund those researchers who have the kind of bias you like, then that full picture might be skewed (this is called publication bias: http://en.wikipedia.org/wiki/Publication_bias).

In this case people worry if the coffee research institute has hired biased researchers, rewards only the researchers who publish good things about coffee or even publishes selectively.

So it is a good thing to point out. And the argument is an attack on the reliability of the fundament for the argument, not necessarily on the researcher.