In addition to things already mentioned....
Another reason for creating the Polyline entity type back when they first came into the picture was to save memory. (Memory was expensive in those days, but it doesn't matter so much today.) A Polyline holds information about its Layer, linetype, linetype scale, color, thickness, lineweight, and so on, once for the whole Polyline, no matter how many segments it contains. The same construction, using independent Lines and Arcs, would need to keep track of all that information separately for every entity, and would thereby consume more memory.
Polylines with non-continuous linetypes can (optionally) have the generation of the linetype pattern flow around the corners / across the vertices, so that the linetype will be apparent even on something made up (entirely or in part) of segments that are too short for the linetype to show if they were independent Lines and Arcs.
Suppose you have an outline of some kind made up of Lines, and you want all the corners rounded. You would need to Fillet each corner separately, picking on two Lines for each corner. With a Polyline, you can use the Polyline option in the Fillet command, and it will round all the corners (details of geometry permitting) with just one pick anywhere on the Polyline.