My assumption is that it’s about risk management for them. Maybe I’m wrong though. Keeping sats locked up in fixed-size channels creates potential risk for the node operators. In their old model, if the sats were sitting on your side of your channel or channels (aka, your wallet balance), then the risk is yours to take and they probably wouldn’t care. But if the sats are sitting on ACINQ’s side of the channel (so you have inbound liquidity), then they’re assuming any risk associated with keeping those funds locked up until you receive a payment that moves them to your side of the channel. By splicing, now they can change the size of your single channel to minimize the amount of sats sitting locked up on their node. When you send and receive, they just adjust the channel size accordingly and you pay whatever mining fee is associated with making that happen. 
Discussion
I see. They should make splicing-in a default setting, but they should make splicing-out only for cases of emergency. I’m not sure what risk they are taking, it seems like if what you’re saying is right they are being stingy and just don’t want to deploy adequate capital. I personally would happily pay high fees (even 1-2%) if a business would maintain the capacity of my channel even if it’s mostly on their side.
I would probably still use it if they did that and the channel size stayed the same until a splice in was needed to increase it. That makes more sense to me from a user standpoint. But there are other good options. I’ve been really happy with Breez for a long time.
I think it’s a great feature if implemented correctly. But I’ve only ever personally used Zeus or my own LND node.