Agreed with the above two, that the problem is in what some people think is implied by the term "anchor."

If you think the anchor is a "starting place" on which the regex algorithm should work, then you will definitely get confused about /\s+$/ and other examples you've shown. The ability of some regex parsers to get a performance enhancement by noticing that the regex ends with an easy-to-find constraint has nothing to do with the syntax of the forward-matching operation.

Maybe this is a bit whimsical, but I've never thought of an anchor as the starting place for traversing a ship, so maybe this is why I've never been confused about anchors with regards to reverse-processing of regular expressions. Only the crabs and rats would consider an anchor as an entrypoint. The sailors surely think of an anchor as a restraint on the usual forward-moving operation of the ship.