A service provider runs OSPFv2 across a metro Ethernet segment connecting two PE routers. The default OSPF dead interval means link failures take up to 40 seconds to detect, causing traffic black-holing during transport failures where the physical interface stays up (an intermediate switch fails). The team needs sub-second failure detection without lowering the OSPF hello/dead timers to aggressive values that would increase CPU load and risk false adjacency flaps. Which solution best meets these requirements?