The formula is: nMinFee = (1 + (int64)nBytes / 1000) * nBaseFee;
It's not about how many receiving addresses there are, but it's the size of the transaction that matters. These huge transactions take up ~49kb. In order for a client to relay these transactions, the fees must be at least (1 + 49,000 / 1000) * 0.02 ltc = 1 ltc. So he's paying 1 ltc on these transaction. And clients will propagate these transactions because there are enough fees. And miners that have not upgraded will write these transactions into blocks.
EDIT: If we increased transaction relay fees to the same 0.1 ltc as transaction fees, then he will need to spend 5 ltc as fees for these transactions. The cost to bloat the chain by 1gb will increase to $1,320... it's more but still fairly cheap.
I don't really understand the inner workings of those transactions (and I'm just discovering about relay fees), but I guess if there was a way to make the transaction fee work per receiving address (like, sending 0.000001 LTC 100 times to different addresses or even from the same address to the same address = paying 100*0.1=10 LTC in transaction fees) this would impair his ability to cheaply make large transactions while not being an issue for legitimate users.
I guess BTC-e caused their own problems by messing with the code and their transaction lower-limit.
But did anyone else have issues? (other than the large txn's and thus the large block-chain and large wallet)
Nope, I'm not having issues, I just want to prevent the block chain from unnecessarily increasing its disk usage. But everything's working fine otherwise.