target-size rule should support position:sticky

Currently the `target-size` rule treats elements with `position: stricky` as though they have position:relative. This could in theory result in false positives when the page is scrolled. I think the best solve for this would be to try and work out the original position of the sticky element and use that as its place in `axe._grid`, rather than its current position.

In this, we may also want to revisit how `position: fixed` elements are calculated. Those too have this issue. We have exception code in color-contrast and target-size to try and avoid false positives from them, but I think there's probably a better / more reliable solution if we adjusted how the grid worked.