Stable version including data availability

This tag is the stable version of the soda-curation package to extract the following information of papers using OpenAI

• XML manifest and structure
• Figure legends
• Figure panels
• Associate each figure panel to the corresponding caption test
• Associate source data at a panel level
• Extraction of the data availability section
• Includes model benchmarking on ten annotated ground truth manuscripts