They aren't waypoints by themselves, but they are sure a useful component if you're building waypoints.

In simplest terms, a signpost is defined by a set of goal locations, and has the ability, given any location in the map, to tell the cost of the lowest-cost path from the given location to any of the set of goals.

All the implementation, storage, map, efficient calculation algorithms, cached values, etc, is "detail" at some high level of algorithmic abstraction, which is why 'signposts' and 'dijkstra maps' are sort-of the same thing. The absolute requirement is that they can tell you the cost of the lowest-cost path from any starting point on the map to a goal.

You can use this in pathfinding your way from anywhere to the goal, simply by choosing your next available step based on whether its location has a lower path cost than the one you're on. The most efficient path is the one that chooses the lowest available path cost at every step. You can also use this the other direction pathfinding from the goal to any particular location, just by calculating a path from that location to the signpost and then reversing it.

If you want to implement waypoints in terms of them, that's fairly easy; you just set a signpost at each waypoint, make sure it has cached path-costs for all the steps on optimum paths to the nearest other connected waypoints, and then you can do your waypoint navigation using graph theory at a high level - where you say 'Mob 37 needs to get from region A to region K, and that involves going via waypoints W3, W8, W9, and W4...." Then you'd use the signpost at W3 to tell the mob how to get from wherever it is in region A to waypoint W3, the signpost at W8 to tell it how to get from W3 to W8, the signpost at W4 to tell it how to get from W8 to W4, and then use W4 again with the reversal trick to calculate a path from W4 to its final destination in region K.

So your long-range pathfinding can be done on a very simplified graph; the waypoints where paths can diverge are the only vertexes you need to consider, and of course they know the path costs to the locations of the vertexes they're connected to. So while doing long-range route planning you don't need to consider each square along the way.