It's actually nice that we can d'n'd images directly on TCP but it makes it a bit fiddly when wanting to position exactly at start of timeline instead because there's no 'auto-stop/-snap' as with audio items.
(demonstration may be exaggerated to get the point across..)
edit:
To clarify, this is only when d'n'd images from outside, once on the timeline they do stop at arrange beginning.
edit2:
Ok, it snaps if grid snapping is enabled. Though I usually have it off when doing slideshows. :P