This is just my point of view, but the purpose of "coyote time" isn't to be a handicap for the player; it's to correct for a natural bias that occurs in platformer design.
The player tries to jump at the last possible moment on the ledge in order to achieve the maximum jumping distance (or sometimes to avoid bonking, etc.). All players have some error in timing, which can be either early or late compared to the point they're aiming for. Valid areas to jump however exist only before the ideal jump point, not after. That means regardless of player skill level, players will fail approx. 50% of the time when aiming for that point, even if they felt like they were on time. In other words, the valid jump area is biased in one direction, while player timing errors extend bi-directionally, which creates the misalignment. Coyote time is a means to correct for this.
Naturally the player can learn from failures and attempt to compensate by jumping sooner. But doing so runs counter to the pressure the game may put on you to jump as late as possible, and I think it's often not easy to determine how much sooner a jump needs to be made.
The window for coyote time could be different lengths of course. Generally it's not meant to be noticeable, but it seems reasonable enough to me that if you wanted a game to be more skillful you could use a short, tight window, while a more casually oriented game could use a longer one.
