Don't vary cache on URL protocol or image size

The protocol is ininfluent, the size also seems to be so, and this way
we can prefetch the next image for on-wiki requests.

Bug: T286356

src/Controller/OcrController.php

@@ -225,7 +225,7 @@ private function getText(): string
         $cacheKey = md5(implode(
             '|',
             [
-                $this->imageUrl,
+                self::transformImageURLForCacheKey($this->imageUrl),
                 static::$params['engine'],
                 implode('|', static::$params['langs']),
                 static::$params['psm'],
@@ -237,4 +237,19 @@ private function getText(): string
             return $this->engine->getText($this->imageUrl, static::$params['langs']);
         });
     }
+
+    /**
+     * Make an image URL suitable to be used as a cache key (e.g. strip protocol)
+     * @param string $url
+     * @return string
+     */
+    private static function transformImageURLForCacheKey(string $url): string
+    {
+        // Strip size, see T286356.
+        return preg_replace(
+            '/(page\d+-)\d+px/',
+            '$1',
+            preg_replace('/^https?:/i', '', $url)
+        );
+    }
 }