Added KaggleERN dataset

[WubbzyFromWuzzleburg]

Contributor: Elliott Huang (elliott500800@gmail.com)

Contribution Type: New Dataset + Documentation + Example Script

Description:
This PR adds support for the KaggleERN (INRIA BCI Challenge) EEG dataset in PyHealth. It introduces a KaggleERNDataset class that validates the expected raw folder structure and provides an offline preprocessing utility to convert raw EEG CSV files into fixed-length epoch/window pickle files for downstream training (keeping the pickle schema compatible with the provided fine-tuning workflow). This PR also updates the API documentation to include the new dataset and adds an end-to-end fine-tuning example for EEGPT on KaggleERN, including instructions for downloading and placing the pretrained EEGPT checkpoint.

Files to Review:

• pyhealth/datasets/kaggleern.py — KaggleERN dataset class + offline preprocessing to window pickle format
• pyhealth/datasets/configs/kaggleern.yaml — Dataset config entry (tables/attributes schema)
• pyhealth/datasets/__init__.py — Expose KaggleERNDataset in the datasets namespace
• examples/kaggleern_finetune_EEGPT.py — Fine-tuning example script (paths are placeholders; includes pretrained checkpoint notes)
• docs/api/datasets.rst — Register KaggleERN in dataset API docs index
• docs/api/datasets/pyhealth.datasets.KaggleERNDataset.rst — New API doc page for pyhealth.datasets.KaggleERNDataset
• tests/core/test_kaggleern.py — Unit tests for dataset verification and optional preprocessing integration

[Logiquo]

LGTM, but probably need to update the code to support the newest changes.

[Logiquo]

The test failed with error TypeError: KaggleERNPreprocessConfig.__init__() got an unexpected keyword argument 'pipeline'