I have no comment on any of the community-aspect of a site like this however the "security" features you are featuring here raise concern from me.
For one, there is nothing to indicate to anyone outside the operation or administration of the site that proper IP masking is taking place. Going to a php page and seeing a printed IP is not proof or confirmation of anything of the sort, given you can just write a very quick script to print any valid IP you want from a list. I would NEVER trust any operators word that my public IP is properly protected, and nor should anyone here, given the complete lack of transparency on this. You should trust your own means of protection, ie using a VPN or proxy before accessing the site.
Also, I don't understand how you think removing a properly implemented hashing function (bcrypt) and replacing it with something else makes me feel any better. bcrypt, when used properly is more than sufficient to protect against db leaks. You are free to modify the software however you wish but coupled with your pseudo-security posturing with the "IP encryption" technology doesn't lead me to believe that it was done with expertese.
Be very careful with the claims being made by this website in regards to security, it should be assumed that it a) doesn't work, or b) doesn't exist.
I agree with what you're saying about the IP masking, without viewing the source code or having access to IPs it's impossible to tell how it works which means you can't tell if it works well at all but if you did have access that'd give you a starting point into breaking the code apart and nullifying it so I think they're kind of stuck there. At a minimum it's a least slightly safer than having your real IP printed on every post.
As for replacing the hashing function, assuming it was implemented correctly, argon2 is very strong although it hasn't been put through the kind of testing Bcrypt has. The main gain is that Argon hits GPU cracking speed harder than Bcrypt does forcing you to spend a lot more money to get the same kind of speeds you would against a bcrypt hash. Both are very slow to crack, nearing impossible against all but the weakest passwords but I like the idea of not using the default hashing bundled with IPB 4 mainly because of how slow Invision were to upgrade once their MD5+Salt hashing was made redundant.
I completely agree you have to take security claims like this at face value but I respect the attempt to address an issue that's plagued the RS community for many years now whether or not the idea that people sell/leak your information is blown out of proportion and whether or not the measures taken by Sharkbrew actually work in the first place.