dicom_ocr.py - fix getting regions from Ultrasound

[howff]

• Refactor the function read_DicomRect_list_from_region_tags which appears in two programs, move it into the library
• Fix the return values to become DicomRectText so they can have an ocrsource of UltrasoundRegion not easyocr
• Consider whether to allow the extraction of ultrasound region rectangles without having to do any OCR
• Consider whether a false-positive huge-font text rectangle found by OCR should be silently dropped (current behaviour), logged, or returned but with ner_engine=hugefont and is_sensitive=0 so subsequent processing can decide whether or not to drop it from the list of rectangles.

[howff]

Recent commits have refactored the code to fix the first two tasks.