I see both sides of this, but still err to the "train and practice" rather than "avoid" approach. I really doubt that planned training spins, conducted with the appropriate cautions, in appropriate type ,airworthy aircraft are a source of great hazard. Yes, loosing control of a twin while doing single engine training can become an immense hazard - but that is not intentional spin training! The "avoid" approach seems to me to be akin to saying that if someone is taught to land really well, teaching a bouce recovery or an overshoot after touchdown skill is not really necessary. Ideally yes, though practically a miss, in my opinion.
I have never accidentally spun an aircraft, other than during very abused handling testing, where a spin was very possible, and the recovery preplanned. I think however, that a mismanaged wing drop stall, which is presumed to be a spin, but is not (yet) tends to strike fear into pilots.
Apparently to me, no matter how vigourously pilots are trained to not stall, they still seem to. As long as a stall, however unintended, happens, spin recovery could rapidly become a valuable skill. Yes, unintended stalls and spins seem to be most common close to the ground, where all is lost (hence my very real resistance to the turnback after engine failure on departure - no matter how many posters say it could be attempted).
The spin training includes as a great value the approach to the stall spin, which itself is important in teaching a pilot what it's like when you're getting close.
The more training one has in the unusual corners of flying, the less alarming those corners will be, and the better natural reactions will be there to get out - or better yet, avoid!