updating pattern clustering

[nkempynck]

Updating pattern clustering in the standard enhancer code analysis.

Faster tomtoms: not one vs one but one vs collection
Dual direction tomtom calculation
Updated default thresholds

To do: update notebook

[nkempynck]

still WIP

[nkempynck]

New feature: x_shift parameter for contribution scores plotting
Added an x_shift parameter to contribution_scores() in src/crested/pl/explain/_contribution_scores.py that allows shifting the viewing window left or right when zooming into specific subsets of a region. Includes bounds checking with a clear error message.

Bug fix: duplicate kwargs in clustermap_tf_motif
Fixed a bug in src/crested/pl/modisco/_modisco.py where aspect, cmap, and norm were passed both as explicit arguments and through **imshow_kws, causing a TypeError. All defaults are now handled through the imshow_kws dict. Also added a cbar_pad parameter for controlling colorbar positioning, and improved colorbar rendering with manual positioning, a custom light-centered colormap, and better label padding.

Modisco pattern matching improvements (_tfmodisco.py)

Changed default sim_threshold from 0.5 to 7.0 in match_to_patterns() and from 3.0 to 6.0 in process_patterns().
Changed default trim_ic_threshold from 0.05 to 0.025 in process_patterns().
Vectorized the pattern matching logic in match_to_patterns() using matrix operations instead of a loop, taking the element-wise maximum of both comparison orders to handle asymmetric similarity scores.
Added a new post_hoc_merging() function that iteratively merges similar patterns above a similarity threshold, with IC-based filtering and optional verbose output. The old implementation is preserved as post_hoc_merging_old().
Added a return_info parameter to post_hoc_merging() to optionally return merge metadata.

[nkempynck]

Also enhancer_code_analysis update

[nkempynck]

notebook that is