Hannibal

It's looks like SB is committed when he leads into two players like this, so we need to invest 21,6 (SB's remaing stack) to win 53,4 (21,6 + 21,6 + 10,20) that would be ~40,5%

Against a range of sets, top two and T98x we have 38,7%. That increases a tiny bit when most likely two (non-diamond) aces are dead. (folded by BB)

This board should hit pretty hard his 3-bet-calling range, especially when we block the two aces and one king. How about just calling the turn and keeping his 8764, TT87 etc. bluffs in the pot? Against his value range we are doing pretty bad (something like 25%), but add some bluffs in there and we have well over 30%

I would slightly adjust my opening range by adding more "bad" suited aces and dry big pairs and removing some of the worst medium/low rundowns.

fold SB/BB to BTN steal is quite useful and u get sample size pretty fast.

"So what are we doing now."
Maybe not fake jamming anymore?

"I can fold out some dry overpairs and this board" - I highly doubt that these guys (76/32 & 58/27) fold any overpairs here. 

Not all fish are equal. They are bad, but in different ways. Try to make good notes and attack their individual leaks.