I am fairly sure this is because, in many abugida scripts, the dominant writing material was palm leaf and the tool was a sharpened point or curved scalpel-like knife rather than a pen. If you curve inward from both sides, you don't have to contort your hand in a 180 degree pattern, which runs the risk of tearing the palm leaf. This is also why the shapes are all curves instead of hard 90 degree angles.

I've read that theory, too, but ဂ is written with just one stroke and if you look at palm-leaf manuscripts in general https://en.wikipedia.org/wiki/Palm-leaf_manuscript there are plenty of angled shapes. (Ironically, the example Wikipedia uses for Burmese has a more square style instead of the now-dominant curves.) So I wonder whether it's all just a post-hoc rationalization for historical happenstance. It would be interesting to know whether e.g. Sundanese ᮕ is similarly written with two strokes.