Add multi-procedure batch processing mode

[dreamlessx]

Description

Add a batch processing mode that can apply multiple procedures to the same image in a single run, generating comparison grids.

Use Case

Surgeons often want to compare multiple procedure options side-by-side. Currently each procedure requires a separate inference call.

Proposed Changes

1. Add batch_procedures() method to LandmarkDiffPipeline that takes a list of procedures
2. Generate all results in a single pass (share landmark detection)
3. Create a comparison grid image showing input + all procedure results
4. Add --procedures flag to CLI accepting comma-separated list

Example

```bash
landmarkdiff infer photo.jpg --procedures rhinoplasty,blepharoplasty,mentoplasty --output comparison.png
```

Acceptance Criteria

• batch_procedures(image, ["rhino", "bleph"]) returns list of results
• Landmark detection is shared (not repeated per procedure)
• Comparison grid image is generated with labels
• CLI supports --procedures comma-separated flag

[P-r-e-m-i-u-m]

"Hi @dreamlessx, I'd like to work on this! I'll implement batch_procedures(), the shared landmark detection, comparison grid, and CLI flag."

[dreamlessx]

Go for it @P-r-e-m-i-u-m! Assigning this to you.

A few pointers for the implementation:

• The shared landmark detection should use landmarkdiff.landmarks.FaceLandmarks — extract once and pass to each procedure
• For the comparison grid, take a look at scripts/generate_comparison_grid.py for the existing grid layout code
• The CLI flag could be something like landmarkdiff infer photo.png --procedure rhinoplasty,blepharoplasty (comma-separated)

Let me know if you have questions!