chore(evidence): stop committing the leave-facing-out catalogs

cat_full's *_shelf.jpg views are crops of add/image6.jpg itself - template
matching puts each one back at 0.998-0.999 confidence - so committing them
reintroduced the evaluation photo as a catalog, which is the leak the audit
removed from stock-lopia. cat_pack is the same views without the shelf copies.

Both are regenerated by lfo_setup.py from stock-lopia and the labels, so the
evidence stays reproducible without the repository carrying the leak. Verified
after removal: lfo_setup.py rebuilds all 16 views.

recognization/docs/research/2026-09-28-recognition-architecture/evidence/.gitignore

@@ -0,0 +1,6 @@
+# Leave-facing-out catalogs are built by lfo_setup.py, not committed.
+# cat_full holds crops of add/image6.jpg itself (template match 0.998+), so
+# committing it would put the evaluation photo back into the repository as a
+# catalog - the exact leak the audit removed from stock-lopia.
+cat_full/
+cat_pack/