The Hines article that this article is written about is open access, and worth a read, especially because the Introduction and Discussion sections look to be fairly comprehensible for people outside the field:
Mice float. All the forced swimming test does is measure the time it takes for them to stop paddling.
As far as "torturing" goes, in the US academic studies require submission of protocols to an Institutional Animal Care and Use Committee with proof that any harm caused to an animal is necessary for the study and meets federal guidelines for animal care (and because funding tends to come from federal sources, there is good incentive to meet these requirements). For studies involving pain, nerve regrowth after an insult, or other similar lines of research, injury may be unavoidable in asking the research question, but how the injury is applied must be approved by these committees. Protocols that involve harming an animal tend to be the most stringently regulated by these bodies, even more so than protocols requiring euthanasia. Furthermore, it is usually in the interest of the scientists performing the research to treat their animals consistently well, lest they find themselves with some inexplicable variables to account for.
Now, whether all or any of that merely constitutes "permitted" torture for our own benefit is a different question, but we do try to ensure that animals are treated as well as possible in a given research context.
http://www.nature.com/tp/journal/v3/n1/full/tp2012136a.html