So, I don't personally think that the reason your transaction wasn't relayed/mined was because of miners refusing to mine the transaction because it was a Satoshi Dice transaction. (Although I can't say that I know the real reason, or I would have posted an answer there.) The major mining pools right now deny this "transaction discrimination" (see this bitcointalk thread). And looking through the blockchain seems to verify this: there are lots of transactions getting confirmed for the well known Satoshi Dice addresses.
I think the reasoning for accepting them is the same reason you state: that if the transactions are paying the mining fee, they are essentially paying their own way. Some may (and do) disagree, but I this is certainly one of the reasons that transactions fees are part of the protocol. As long as there is a cost to send a transaction (even if it is a small one) then that does protect the network from spam and DDoS attacks.
As far as the second part of your question "Is it really okay in the Bitcoin network to refuse transactions based on personal preference of the operators (aka Miners), in kind of like the censorship style that many proponents of BTC aim to avoid?", that is somewhat different.
Yes, it is entirely within the protocol (and I would assert the spirit) of the protocol. Bitcoin to a large extent a democracy (except that it is isn't one vote one person). And if there was an attempt to subvert the system, then it is the miners who collectively protect against that.
For example, the BTC market is now large enough that it is conceivable that someone with a vested interest in the BTC exchange rate might find it worthwhile to flood the network in an attempt to devalue BTC, even if it required paying transaction fees on that flood. Another example, if someone found an exploit (or even simple bug in the Satoshi client) that allowed some sort of malformed transaction to crash clients. In most cases, it is entirely reasonable that the miners (from their own interest in protecting the network) would refuse to mine those transactions.
So, yes, I would say that it is entirely reasonable for miners to refuse to mine transactions based on "personal preference". Because that will only affect the network if the majority (actually it likely would take a significant supermajority) were acting on the same personal preference. Which means that they weren't personal preferences so much as a group consensus about the benefit of the network. And since they would be passing up the opportunity to make money on those transactions it is extremely likely that the supermajority could only be reached in the case of a significant network issue. (A great example of this is the March 2013 fork in the network.) The miners collectively chose to ignore the longest blockchain (the 0.8.1 branch) in favor of keeping compatability with 0.7.
