feat(phase1): implement OpenCV image processing pipeline

- Create ImageProcessor service with EXIF orientation detection
- Implement smart cropping via OpenCV contour detection (10% padding)
- Add text orientation detection using Hough line transform
- Resize and compress images to 1200px with 85% JPEG quality
- Generate 200px square thumbnails with center crop
- Fallback to Pillow if OpenCV fails
- Comprehensive test suite: 28 tests all passing
- File size validation (reject >10MB)
- Graceful error handling for corrupted/invalid images
- Update requirements.txt with opencv-python, piexif, python-magic

venv/lib/python3.12/site-packages/piexif/_transplant.py

@@ -0,0 +1,45 @@
+import io
+
+from ._common import *
+
+
+def transplant(exif_src, image, new_file=None):
+    """
+    py:function:: piexif.transplant(filename1, filename2)
+
+    Transplant exif from filename1 to filename2.
+
+    :param str filename1: JPEG
+    :param str filename2: JPEG
+    """
+    if exif_src[0:2] == b"\xff\xd8":
+        src_data = exif_src
+    else:
+        with open(exif_src, 'rb') as f:
+            src_data = f.read()
+    segments = split_into_segments(src_data)
+    exif = get_exif_seg(segments)
+    if exif is None:
+        raise ValueError("not found exif in input")
+
+    output_file = False
+    if image[0:2] == b"\xff\xd8":
+        image_data = image
+    else:
+        with open(image, 'rb') as f:
+            image_data = f.read()
+        output_file = True
+    segments = split_into_segments(image_data)
+    new_data = merge_segments(segments, exif)
+
+    if isinstance(new_file, io.BytesIO):
+        new_file.write(new_data)
+        new_file.seek(0)
+    elif new_file:
+        with open(new_file, "wb+") as f:
+            f.write(new_data)
+    elif output_file:
+        with open(image, "wb+") as f:
+            f.write(new_data)
+    else:
+        raise ValueError("Give a 3rd argument to 'transplant' to output file")