The first app created a layer between twitter and its users which meant that stripping ads would be trivial & likely universal (they don't put ads in 3rd party app streams yet but should they chose to this would stop them) and you could be harvesting the analytics and user metainfo that twitter wants to sell for ad-targeting against what they probably consider "their" content. I can see how they would take exception to a service that allows users to fully control their interaction with twitter; it could seem too much like "handing the keys over" to a third party.
TweetFavor (ironically) seems that it would worsen the signal to noise ratio that Proxlet aimed to improve. Friends turning themselves into voluntary zombies in a twitter spam botnet? I know it's not actually that bad but I can see why twitter would want to cut this off.
None of this excuses the manner with which twitter did what they did, and like I said, "Devil's Advocate," neither of the apps were that bad, but it's not unfathomable to me why twitter would have a problem with them.
TweetFavor just supports an existing pattern "hey dude, can you tweet this for me?" And permission had to be given per-tweet request, there was no mass zombification. You'd pledge individual tweets. Once again, no reason as to why it was shut down, and certainly no one used it before making the call.