[BUG]: `Image\Adapter\Imagick::processReflection()` advances the wrong frame cursor

[niden]

Background

The max-level pass over Image surfaced two defects. The second one - processText() passing a boolean $offsetY to annotateImage() - has been fixed and is recorded at the bottom of this document for history. This one remains.

The defect

The first two loops in processReflection() mutate $reflection but use $this->image->nextImage() as the loop condition. The two objects hold independent frame cursors.

$reflection->setIteratorIndex(0);

while (true) {
    $reflection->flipImage();
    $reflection->cropImage($reflection->getImageWidth(), $height, 0, 0);
    $reflection->setImagePage($reflection->getImageWidth(), $height, 0, 0);

    if (true !== $current->nextImage()) {   // <- advances the wrong object
        break;
    }
}

$reflection's cursor is never advanced, so for an image with N frames:

• Loop 1 applies flipImage()/cropImage() to $reflection frame 0, N times over, instead of once to each frame. An even N cancels the flip entirely. Frames 1..N-1 are never touched.
• Loop 2 then finds $this->image's cursor already exhausted by loop 1 and breaks after a single pass, so the fade gradient composites onto frame 0 only.
• Loop 3 does call setIteratorIndex(0) on the cursor it advances first, so it is correct.

The likely intent is $reflection->nextImage() in both loops, but that must be confirmed against ImageMagick's frame semantics - the two loops may also each need their own setIteratorIndex(0) reset.

Single-frame images are unaffected. With N=1, nextImage() returns false on the first call, every loop runs exactly once, and the output is correct. That is the entire reason the suite is green.

[niden]

Resolved in #17513