This is a well written document that is ready for publication.

There is one point that the IESG should ponder. The authors have asked for a IP type assignment. This is a limited registry that needs to last the lifetime of the IP protocol suite. NSH started its life 9 years ago and has been a standard for 4 years and in all this time has not needed such as allocation. Neither SRv6 nor NSH are petite or lightweight protocols. So I wonder if the identification of NSH should happen at the IP layer as proposed, or whether an intermediate multiplexing layer such as UDP should be used? The extra processing for UDP is one test and the extra MTU is 8 octets. The decision for the IESG is whether in their view the extent of deployment and the gain in performance is such that they should authorise the allocation of the IP type.

One small point, I assume that the IANA allocation in Section 11.2 "SRv6 Endpoint Behaviors" should come from the FCFS series of code points.