1) Should be doable, but I'd not recommend it. Reasoning being, if that one hub goes down, all servers get split from each other, and you'll get one hell of a resynch happening... not fun.
2) Not sure, not had time to look in detail sorry.
3) Everything will split and rejoin, massive bandwidth use, etc. -- see 1.
4) Nope [and why are you going to need more, clients will get sendq'd requesting the /names for a channel as it is, trust me, i know ;P]
dammit. This thing costs a lot of money. It isn't worth to spend it if the purpose of this toy is just a botnet. Also a great charge can (and probably will) be for such thing.
show me some reasonable (!) arguments how someone may build up an over-1Mio-User-net just from scratch, while networks having grown up vor several years are stumbling @ ~200-240k at most
Tbh, i say either botnet or 'wannabee attempt for global known/reigning irc network from scratch'.
Either way not gonna happen imo...
Edit: Either above or the dude is loaded an bought the major three players worldwide an plans to move them to his own layout. Unlikely. Conclusion remains.