Ah, well if you can run two IOS instances in software with 8 GB, that's an option then, but RPR and NSF vs. HSRP/GLBP aren't mutually exclusive, as they serve different purposes.
The second IOS image is to deal with a major IOS fault, such that the second image can take over. This isn't something I don't believe is seen very often, so it's unclear how much value this offers, but if the equipment has the capability, like chicken soup, it probably wouldn't hurt to enable it. I.e. normally, for such redundancy, there's a second hardware "brain" to take over.
NSF keeps forwarding packets after higher logical levels (in the stack) have failed and during their recovery. I haven't read what exactly an ASR1001-X support, but it may, in the case of switching to its second/backup IOS process might keep forwarding packets and not "drop" as far as NSF aware neighbors are concerned. (The latter might be an issue with your WAN side, as the external WAN devices, might not be NSF aware.)
Something like HSRP/GLBP might benefit from RPR and NSF too, as they too might not see the peer "drop".
