And health benefits of yoga are probably not strongly correlated to gender.

This is why it surprises me that there is a gender imbalance in people going to yoga classes.

Yet people engage in leisure activities generally not because of possible long-term benefits, but because they find these activivities intrinsicaly rewarding. And since different people have different preferences, the activities they find rewarding differ. Add the fact that leisure time is a limited resource, hence a tradeoff between available activities must be done, and these competing activities, and their enjoyablility differs from one person to the other.

Personal preferences correlate with gender.

Here, again, I think that a large part of the difference between personal preferences with gender is more cultural than biological. Consider, for example; culturally, over a large part of the world, it is considered acceptable for a woman to wear a skirt, but frowned on for a man. As a result, few men wear skirts; if you were to pick a random man and ask for his opinion on wearing a skirt, it is likely that he would not wish to do so. However, if one considers a slightly different culture for a little (for example, the Scottish kilt), one finds a similar garment being worn by many men. So a person's preferences are affected by culture.

But I don't think the nature vs nurture question really matters here: different people have different preferences, whatever the cause, and I don't see why we should try to engineer them to achieve some arbitrary ideal.

I don't see it so much as reaching an arbitrary ideal; I see it more as avoiding a known failure mode.

I have noticed that, throughout history, there have been cases where people were divided into separate groups; whether by race, gender, religion, or other means. In most of those cases, one group managed to achieve some measure of power over all the other groups; and then used that measure of power to oppress all the other groups, whether overtly or not.

This leads to all sorts of problems.

One means of maintaining such a division, is by creating a further, artificial divide, and using that to widen the gap between the groups. For example, if significantly more men than women own land in a given society, then restricting the ability to vote to landowners will tend to exacerbate any official pro-male bias. (This works the other way around, as well).

Therefore, when I see a major statistical imbalance for no adequately explained reason (such as the noted gender bias on LessWrong, or the imbalance in yoga classes) I find it a cause for slight concern; enough to at least justify trying to find and explain the reason for the imbalance.