splitting signatures from the UTXO merkle tree hashing will cause big problems
You should take a moment and go look at how SegWit actually works.
It sounds like the thing driving your fear of it is based on a false assumption about how it works.
If the signatures ARE committed to the merkle tree hashing of the block (and therefore committed to the merkle root), does that put your mind at ease and make you feel comfortable that SegWit does NOT create a situation where miners could just ignore signatures?