Generate metrics for comparison between tesseract and tr-ocr

[jonchang]

Metrics were presented as part of IDWA RV/RL biweekly shareout meeting on 12/5. Raw data are also available on google drive here - https://drive.google.com/drive/folders/1nDBJl4pyPjee8Sri2rFjt2xedHXOUwOH?usp=drive_link

[jonchang]

Current status:

• scripts and other backend code are being updated in Update scripts for metrics computation #443 to handle additional metrics computation for the new tesseract backend
• an issue was discovered with the benchmarking segmentation template that led to certain fields being inadvertently truncated. this caused the accuracy of the models to artificially decrease as the ground truth data was not affected. all benchmarks are currently being re-run to get more correct accuracy information for tr-ocr and tesseract backends.

[jonchang]

Outstanding items:

• use tessdata_best in Use better tesseract training dataset #459
• upload new data and fixed segmentation template to google drive folder

[jonchang]

Follow up - update report with new metrics - https://docs.google.com/document/d/17KV8ehzcvXtToibHuO_NhblzRz1Miz5BiSe3Nrj8WHI/edit?tab=t.0