Fix image dimensions in detections stitch block output

[brunopicinin]

Description

• Fix Detections Stitch block to output correct image dimensions in detection metadata. Previously, dimensions were being reused from existing values in sv.Detections, which had metadata from crops' image sizes.
• Use util function to update all parent metadata properly
• Set proper mask resolution to fix stitching masks when single block step testing

Before fix (detections "outside" of image):

CleanShot.2025-10-29.at.15.14.44.mp4

After fix:

CleanShot.2025-10-29.at.15.13.43.mp4

Type of change

• Bug fix (non-breaking change which fixes an issue)

How has this change been tested, please provide a testcase or example of how you tested the change?

Locally

Will the change affect Universe? If so was this change tested in universe?

No

Any specific deployment considerations

No

[codeflash-ai]

⚡️ Codeflash found optimizations for this PR

📄 274% (2.74x) speedup for manage_crops_metadata in inference/core/workflows/core_steps/fusion/detections_stitch/v1.py

⏱️ Runtime : 1.33 milliseconds → 356 microseconds (best of 385 runs)

A dependent PR with the suggested changes has been created. Please review:

• ⚡️ Speed up function manage_crops_metadata by 274% in PR #1654 (fix/stitch-block-image-dimensions) #1655

If you approve, it will be merged into this PR (branch fix/stitch-block-image-dimensions).