Keep every packet label when a side is photographed again, and use the newest
photo of a side. Before, giving a second photo the same label cleared the
older photo's, which let it back into the crop's photos and default picture.
Carry the label through the HTML photo picker (its search form and the links
that add a photo), so a packet photo added without JavaScript is labelled.
Have the OCR script treat any network failure on an image (DNS, timeout) as
that image failing, so the rest of the batch still runs.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
- Introduced a Rake task `packets:needing_ocr` to list photos of seed packets that lack text, outputting results as JSON lines.
- Created a Python script `read_packets.py` to process the output of the Rake task, downloading images and performing OCR to extract text.
- Updated model specs for `Crop` to ensure proper handling of photos associated with different seeds.
- Refactored API request specs for `Photos` and `Seeds`, removing unnecessary attributes from the response.
- Added tests for the new Rake task to ensure correct functionality and output format.
