Fixing duplication issues in few samples.

Removing more coordinates from data.

Adding new iteration of the dataset files.

Adding intermediate format, where JSONL is present, but we also use our list-based sentence structure for each text.

Changing JSON format to JSONL.

Adding initial data release.