The trouble with bigger-is-better as applied to color gamuts is that some of your "bit budget" is getting "spent" representing colors that
aren't reachable, instead of colors that
are reachable.
The article I cited in my earlier post has a diagram showing this (in the "Influence on Bit Depth Distribution" section, about halfway down.)
That's not a problem when working with 16-bit-color files, since there are plenty of bits to spare. But in 8-bit color, this can result in problems such as banding and less-smooth color gradation. Since 8-bit color is the norm for displays and even for most printer drivers, this is an issue that can't just be ignored.
So yes, for your 16-bit master files, you'll want to choose a color space with as wide a gamut as possible, such as AdobeRGB (which is optimized for 4-color press printing, by the way) or ProPhoto RGB (which is optimized for photographic printing.)
For files you'll reduce to 8 bits, you have to do a bit of thinking about where they're going. I spend every workday making decisions of this sort, and I pick sRGB if I'm sending the file into a non-color-managed environment (office users) or AdobeRGB if I'm sending it into a color-managed environment (professional graphics users.)
See, I
warned you we could wind up drowning in theory...