Flaky feature extraction tests for Donut

### System Info Note: I have observed with my own setup, but most recently this was seen in a CI environment - `transformers` version: 4.26.0.dev0 - Platform: macOS-13.0.1-arm64-arm-64bit - Python version: 3.9.15 - Huggingface_hub version: 0.11.0 - PyTorch version (GPU?): 1.14.0.dev20221118 (False) - Tensorflow version (GPU?): 2.10.0 (True) - Flax version (CPU?/GPU?/TPU?): 0.5.3 (cpu) - Jax version: 0.3.6 - JaxLib version: 0.3.5 - Using GPU in script?: Noe - Using distributed or parallel set-up in script?: No ### Who can help? @amyeroberts ### Information - [X] The official example scripts - [ ] My own modified scripts ### Tasks - [X] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...) - [ ] My own task or dataset (give details below) ### Reproduction pytest tests/models/donut/test_feature_extraction_donut.py::DonutFeatureExtractionTest ### Expected behavior Tests don't randomly fail e.g. like in [this CI run](https://app.circleci.com/pipelines/github/huggingface/transformers/53575/workflows/40f5b896-c941-4a9f-8946-26f54bb14505/jobs/644204). There is an issue occurring when the amount to pad is being calculated - some dimensions are having negative padding found.