Okay, I checked the values for an o-ring cross section of 0.139, a groove width of 0.136 and a groove height of 0.100. (By the way, all those values are imperial (inch), right?)
If those values are correct and assuming no tolerances, then the fill rate is at 110%. And as I said, while rubber can change shape under compression, it can not change in volume. So either the housing will not go into contact (or at least bends), leaving a gap just big enough for the rubber material to squeeze in, or something gets damaged. Either is not exactly good. You would not want the o-ring to squeeze/bend over the edge of the groove.
Squeeze (compression) would be within recommendation at 28%. A fill rate of 100% would be achieved at a height of 0.110, and at 0.115 height you'd have 95% fill rate with a squeeze of 17%, which would still be within recommendation.
And you are right, you can intentionally make a groove narrower than the cross section of the o-ring for it to not drop out during assembly, however that is not good design. You need squeeze for a seal, and you need space for the rubber to move to, no way around it. A design with a groove narrower than the o-ring cross section means you can't close the groove, risking damage at the groove edge when under pressure.
But you are right, at the moderate pressures under water plus the ability to easily exchange the o-ring for underwater housings you can probably get away with a lot of "bad design".
If you mean what I think you mean with dovetail design, then yes, that is a better design. The standard is to make the inside diameter of the groove bigger than the inside diameter of the o-ring, so the o-ring stretches and keeps in place that way (which is also the case with my housings, and they don't pop out during assembly).