It's an insane feature for library authors TBH, it opens the door to simplifying so many DSLs (which are extremely heavily used in Kotlin).
I wouldn't say we're going "all-in" in Arrow, but rather we're going "all-in" on DSLs. And as a result that makes context receivers super exciting, but we've already been on this path much longer than we've realised just how beneficial context receivers are for us.
Our DSL process started 3-4 years ago when we wanted to move away from the typeclass/HKT encoding, and it's only been like 1 year since we started playing with context receivers when they experimentally become available on JVM.