One of the points is that settlements systems are private permissioned blockchains, which don't take advantage of one of the main innovations of the public blockchains, i.e. decentralised trust. Interestingly I noticed that the recent Bank of England Real Time Gross Settlement Proof of Concept report[0] didn't even use the term "blockchain", preferring instead to use the term "distributed ledger technology" - I wonder if they are concerned about negative connotations the term "blockchain" might be starting to have.
[0] https://www.bankofengland.co.uk/-/media/boe/files/payments/r...
I'd argue that settlement systems aren't a niche system and is something most of us use on a daily basis -- if not more frequently. Unless you're using cash for everything, every transaction you make goes through a settlement process, sometimes multiple times.
People often talk about how inefficient blockchain is, but they rarely have a grasp on how horribly inefficient our current settlement systems are. They are slow, expensive, inefficient, and complex.
I have doubts if it'll be widely used outside of the financial space, and I recognize that may be the context in which you were using the adjective "niche", but I do think it has the potential to be transformative within that space, and I think that space is potentially bigger than many of us realize.