As long as you only need one frame to be visible at a time, you should keep the frames in one slot or you end up with this mess.
Also as long as you only key one frame to be visible at a time, the number of polygons/triangles wouldn't be the problem.
It's the number of keys the runtime has to crunch through for every timeline and every frame.
I think showing an attachment name (and image) on hovering over the attachment key is planned to come with the upcoming Slot Image Chooser UI.
The way Unity's Animation UI solves this is by putting thumbnails on the dopesheet. It's helpful for drastically different-looking frames but not for relatively high-framerate animations, or gradual animations where the previous and next frames look practically similar in thumbnails.