fix: exclude deprecated entries from existing_data_map lookup

The existing_data_map was including ALL entries, so when there were two
entries with the same full_name (current + deprecated), the last one would
overwrite. After sorting deprecated entries last, this meant deprecated
entries were overwriting current ones in the lookup map.

This caused the scanner to compare against deprecated entries instead of
current ones, incorrectly detecting MD5 'changes' for files that hadn't
actually changed.

Fixed by excluding deprecated entries when building the map - only current
(non-deprecated) entries should be used for file comparison.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Author: devonjones

openforge/data/incremental.py

@@ -65,8 +65,12 @@ def __init__(
         self.verbose = verbose
         self.skip_schema_validation = skip_schema_validation
         self.existing_data = self._load_fixture()
+        # Only map non-deprecated entries - deprecated entries shouldn't be
+        # used for comparison since they represent old versions
         self.existing_data_map = {
-            item["file_metadata"]["full_name"]: item for item in self.existing_data
+            item["file_metadata"]["full_name"]: item
+            for item in self.existing_data
+            if not item.get("deprecated")
         }
         self.subset_path = self._detect_subset_path()