Fix IADDAT visualization anchoring to model coordinates instead of periodic images

[Copilot]

Displacement vectors and density peaks were displayed at shifted unit cell copies rather than around the atomic model. The visualization logic incorrectly used mark.pos (nearest atom image from gemmi.NeighborSearch, which may be a periodic copy) as the anchor point.

Changes

Map Generation

• Use full unit cell map instead of ASU-masked map in IADDAT_peaks_table
• Enables peak finding throughout the cell regardless of model placement

# Before: Limited to asymmetric unit
asu_map = sf.masked_asu()
realmap = np.array(asu_map.grid, copy=False)

# After: Full unit cell
realmap = np.array(sf, copy=False)

Vector Calculation

• Anchor displacement vectors to original atom coordinates (coordx, coordy, coordz) instead of mark coordinates in calculate_per_atom_vectors
• Compute vectors as peak - coord rather than peak - mark

Peak Visualization

• Position peaks at coord + delta relative to model atoms in create_peak_objects_single_model and create_carved_peak_objects_single_model
• Ensures peaks appear superimposed on the model structure

Removed Incorrect Symmetry Logic

• Deleted 8 functions (~430 lines) that misinterpreted image_idx from gemmi.NeighborSearch as space group operations
• image_idx represents PBC shifts, not crystallographic symmetry operations
• Removed: map_symmetry_features_to_model, extract_symmetry_transformations, apply_inverse_transformations_to_peaks, and related validation functions

Original prompt

Fix vector mapping and peak visualization in IADDAT plugin.

Currently, displacement vectors and density peaks may be displayed in shifted unit cell copies rather than around the atomic model. This occurs because the visualization logic uses the coordinates of the "mark" (nearest atom image found by neighbor search, which may be a periodic copy) as the anchor, rather than the original atom coordinates. Additionally, the plugin attempts to map symmetry using image_idx as a space group operation index, which is incorrect for gemmi.NeighborSearch (where it represents PBC shifts).

This PR fixes these issues by:

1. Using the full unit cell map instead of an ASU-masked map in IADDAT_peaks_table. This ensures peaks are found throughout the cell, allowing NeighborSearch to find matches near the model regardless of its placement relative to the standard ASU.
2. Anchoring visualization to the model:
   • Displacement vectors are now drawn starting from the original atom coordinates (coordx, coordy, coordz).
   • Peaks are visualized at positions relative to the model atoms (coord + delta), ensuring they appear superimposed on the model structure.
3. Removing incorrect symmetry mapping logic: The functions map_symmetry_features_to_model, extract_symmetry_transformations, apply_inverse_transformations_to_peaks, and related validation methods are removed. These were based on a misunderstanding of gemmi's image_idx in this context and caused the "shifted unit cell" artifacts.

This ensures that the IADDAT analysis correctly visualizes the difference density and displacement vectors directly on the loaded model, solving the issue of vectors pointing to shifted unit cell copies.

This pull request was created from Copilot chat.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.