Symptom:
As of 5.2.0, RIB is hot standby to support NG-ISSU and NSR. NSF is not impacted and continues to
operate on redundancy switchovers and failovers.

However the following implementation is being documented:

IGP IPFRR backup paths (which are not used for fowarding) are not synchronized to the standby RIB.
On switchover, the IGPs will repopulate the RIB with these backup paths. There is a potential
window where FIB may sweep/purge the IPFRR backup paths temporarily  before the IGPs re-add
the paths to RIB.

If an FRR event was to occur in this window, loss of forwarding will recover based on IGP
convergence.

Conditions:
A double failure condition is required:

1) RPFO event
2) FIB has performed mark-sweep of IGP routes *before* IGPs have re-added the IPFRR paths.
3) A FRR event has to happen in the window 2) above (i.e interface down/flap)

This condition has such low probability that implementation of RIB syncing of IPFRR paths
will be taken as a customer feature request.