Allow dicom files to work with process_image

[ieee8023]

Also ran black formatter on the files.

[Copilot]

Pull Request Overview

Adds support for DICOM files in the image processing pipeline and applies Black-style formatting across utility and script files.

• Implements a 128-byte preamble check for “DICM” in load_image to route DICOM files through read_xray_dcm
• Refactors exception messages and multiline call formatting in torchxrayvision/utils.py
• Updates scripts/process_image.py to use the new load_image function and applies consistent quoting and line breaks

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
torchxrayvision/utils.py	Add DICOM preamble detection in load_image, refactor formatting and exception types
scripts/process_image.py	Replace manual imread/normalize logic with load_image, clean up argument parsing formatting

Comments suppressed due to low confidence (4)

torchxrayvision/utils.py:211

• The key used to track normalization warnings is inconsistent ("norm_check" vs "norm_correct"), causing the check to never be bypassed; consider using a single consistent key.

if not "norm_check" in warning_log:

torchxrayvision/utils.py:81

• The newly added DICOM detection branch in load_image doesn’t appear to have accompanying tests; consider adding unit tests to cover reading .dcm files.

with open(fname, "rb") as f:

scripts/process_image.py:21

• [nitpick] The "-f" argument is defined but not used anywhere in the script; consider removing it to reduce confusion.

parser.add_argument("-f", type=str, default="", help="")

torchxrayvision/utils.py:57

• [nitpick] Replace this generic Exception in normalize with a more specific type like ValueError to make error handling clearer.

raise Exception(