Fix PILResizeCrop pre-resize loop using inconsistent height threshold

[Chessing234]

Bug

`tools/extra/resize_and_crop_images.py:PILResizeCrop.resize_and_crop_image` does a power-of-two pre-shrink loop before the final thumbnail:

```python
#preresize image with factor 2, 4, 8 and fast algorithm
factor = 1
while img.size[0]/factor > 2*box[0] and img.size[1]2/factor > 2box[1]:
factor *=2
```

The two guards are supposed to check the same thing — "both dimensions are more than 2x the target box" — but the height side multiplies by 2 where the width side does not.

Simplified, the effective per-dimension thresholds are:

Dim	Current	Effective
width	`img.size[0]/factor > 2*box[0]`	width > 2·box·factor
height	`img.size[1]2/factor > 2box[1]`	height > 1·box·factor

So `factor` keeps doubling as long as height > 1x target, not 2x. The loop over-shrinks tall/narrow images (factor grows one extra step past the documented 2x safety margin), which then feeds a smaller-than-intended NEAREST thumbnail into the later ANTIALIAS resize at line 67 — hurting output quality relative to the intended behavior and to the sibling width check.

Root cause

The `*2` on `img.size[1]2/factor` is a stray token left over from the source snippet the comment points to (`http://united-coders.com/christian-harms/image-resizing-tips-every-coder-should-know/\`), where both guards are symmetric (`img.size[i]/factor > 2box[i]` for i in 0, 1). The `2` belongs only with the `box[...]` term on the right-hand side, not multiplied into the image dimension on the left.

Why the fix is correct

Dropping the `*2` restores the documented behavior: both dimensions use the same "still more than 2x the target" guard, so the NEAREST pre-shrink stops at the largest power-of-two factor that leaves enough headroom for the final ANTIALIAS resize on line 67. No other call site or behavior changes — only the height branch of this single guard is corrected.