Screenshot names: on-device text reading models (PP-OCRv5 mobile)

These models are not part of the app. They are downloaded from our own
storage the first time a screenshot is named on a device without a built-in
text reader, run only on your device, and never send the image or its text
anywhere.

PaddleOCR PP-OCRv5 mobile text detection and text recognition (multilingual
Chinese/English/Japanese, Latin, Cyrillic, Arabic, Korean, Devanagari, Thai
and Greek), from the PaddlePaddle models on Hugging Face
Copyright (c) PaddlePaddle Authors (Baidu, Inc.).
Licensed under the Apache License, Version 2.0.

Changes made by ITANU STUDIOS LTD: the models were converted from the
PaddlePaddle format to the ONNX format with paddle2onnx, and each
recognition model's character list was saved as a separate text file. The
model weights were not otherwise changed. Each downloaded file is pinned by
its SHA-256 digest.

The Apache License, Version 2.0 follows.
