I've never actually heard of the salt water test. It makes sense thinking about it. I got a bag of dice from my dad, from his d&d days, from when he was younger than me. There's a huge mix of some really weird dice in there. They're fairly worn, some of the numbers aren't visible any more, and their original quality ranges wildly. I wouldn't mind dropping them all in a bucket of salt water to see now. I've drawn my own conclusions about them over the years of using them, it'd be interesting to see how well they match up.
Also, i wonder how my dad's homemade d20, made out of a double d10(d20 with 0-9 twice, 0's used as 10 and 20) with half the numbers coloured red(11-20 is red), with a pen, would rank up, it was likely just done randomly.
That said, I'm not sure that's generalizable to more realistic amounts of bouncing ("only a bit of bouncing" as opposed to "none at all", and "kinda a lot of bouncing" as opposed to "effectively zero resistance").