Multiple reasons:
1)Security is our top priority so in order for the signers to be able to register and validate each deposit 100% reliably, the transaction from the intermediary address to the multi-sig needs to be broadcasted. We are sacrificing some sats paid extra in fees for an unbreakable system in regards to loss of funds for any reason other than us, the operators, acting maliciously.
2)We might not necessarily pay extra in fees since right now we can broadcast all deposit transactions deposit x > multisig on a very low fee regardless of congestion. Output transactions from the multi-sig need to be broadcasted with higher fees so they don't get stuck.
3)We want to leave some UTXO's available in case a transaction still gets stuck even with a higher fee

Other than working on other features and exploring ways to decentralize the service completely our job should now be as easy as this: pay the servers and change them once in a while just in case. Whirlwind could be considered a blockchain, the only missing piece is decentralization meaning more signers. We are considering a system where in order to be a signer you are required to deposit a certain amount of BTC, and we implement a slashing mechanism to deal with bad actors. We will give more details once we have an actual plan, until then we're patiently waiting to see if demand for something like this actually exists. From a technical point of view we are by far the superior privacy solution available for Bitcoin today, so if something like this isn't used (even when it's essentially free) then it's certainly not worth wasting time to develop it further.