[Benchmarking Framework] Establish e2e benchmarking framework

[zdeveloper]

Establish e2e benchmarking framework by using the created reportvision-dataset-1
we want to automate the measurement or accuracy and speed for running that dataset into the ocr pipeline

Acceptance Criteria

• write automated tests to measure e2e OCR reportVision accuracy and time to run against the reportvision-dataset-1
• also measure the confidence accuracy and at what confidence level are the results not correct
• create the templates manually or using frontend, creating the templates should not be part of the benchmark
• store the benchmark with all required data/scripts on a benchmark/reportvision-dataset-1 with clear documentation on how to run it.

Additional context

• for python example tests, look at OCR model benchmark tests

[zdeveloper]

per @schreiaj, please save the segmentation templates so its also standardized for every run

[arinkulshi-skylight]

The steps taken to so far to address this ticket:

• Created segmentation template for datasets, updated output of labels.json file
• Created and implemented batch segmentation and OCR processing script
• Added time measurement to OCR processing script
• Create batch metrics for OCR pipeline
• Create csv of leven + confidence for values where leven is greater than 1.