I'm with spartacusrex. The ultimate test is for someone to pull of one of these (theoretical) attacks and catastrophically and irreparably damage the network in some way, or at least prove that one of the attacks can be used to consistently and successfully attack the network and/or individual users. Until this test is completed, I'm going to assume that POS and other variations (DPOS) is sufficiently secure.
Also, it would be in everyone's best interest if POS was broken sooner rather than later while valuations are low. So please, if you have a guaranteed attack, go ahead and do it and prove POS useless.
I didn't read the whole thread so maybe this is covered but the reason* these attacks don't happen in practice is that none of the
deployed chains actually operate as decentralized consensus systems. They are centralized in some manner with checkpoints, centrally signed blocks, etc.
This makes them impossible to attack but it also makes them a sham. They're just centralized systems implemented in an inefficient way that gives the appearance of decentralization.
* The other reason these systems aren't necessarily attacked is that attacking takes work, and often no one really cares (competent people have better things to do). Shadowcash was recently deanonymized due to a mathematical flaw that rendered their anonymity technique utterly and completely worthless, and which existed in their design for over a year, with a bounty offered. But the flaw was only discovered by accident. Apart from this accidental discovery, the flaw could easily have stayed there for years longer but that would not have made the system any less worthless. Do not assume that since something hasn't been broken yet, it is secure. That is completely wrong.