Bump document sync hash to c95b4f2 and drop the transitional DEVIATION

Two follow-ups after the DID-python#83 / DID-matlab#198 merge landed
on main:

  * did_matlab_python_bridge.yaml: matlab_last_sync_hash for the
    document entry was still 78583f6, so the bridge coverage job
    flagged c95b4f2 (the MATLAB parse-tighten commit) as unreviewed
    drift. Bump it and add a class-level decision_log note naming
    the commit and pointing at the method-level entries that already
    describe the alignment. No Python code change was needed -- the
    Python side was already strict.

  * tests/symmetry/read_artifacts/common/test_series_member_names.py:
    the transitional DEVIATION for chunkdata.bin_-1 (added while
    #198 was pending) is now stale. Both languages refuse the name
    at the parse; empty the table.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SqTZnkGgPBgWKyAyNHGNuv

Author: claude

src/did/did_matlab_python_bridge.yaml

@@ -906,7 +906,7 @@ classes:
   - name: document
     type: class
     matlab_path: "+did/document.m"
-    matlab_last_sync_hash: "78583f6"
+    matlab_last_sync_hash: "c95b4f2"
     status: ported
     python_path: "did/document.py"
     python_class: "Document"
@@ -1371,6 +1371,16 @@ classes:
       deliberately not ported. Synchronized 2026-03-16; construction and
       validation 2026-08-28.
 
+      Sync hash bumped to c95b4f2 on 2026-09-08 (DID-matlab#198 /
+      DID-python#69). That commit tightened the NAME_<i> parse in
+      is_in_file_list, seriesMemberOf and localValidateFileDeclarations
+      from str2num (which EVALUATES) to a strict all-digits check, matching
+      Python's series_member_of. The Python side was already strict, so no
+      code change was needed here -- the entry's method-level decision_logs
+      for is_in_file_list and the constructor already describe the alignment.
+      Both languages now accept NAME_<i> as a numbered entry only when <i>
+      is a non-negative integer.
+
   # -----------------------------------------------------------------------
   # did.query
   # -----------------------------------------------------------------------

tests/symmetry/read_artifacts/common/test_series_member_names.py

@@ -31,23 +31,11 @@
 
 # name -> (this language's stem, this language's index, why they differ)
 #
-# TRANSITIONAL, pending VH-Lab/DID-matlab#198. Once that PR merges,
-# MATLAB's seriesMemberOf uses a strict all-digits parse and this table
-# empties out. Until then the DID-matlab main this workflow checks out
-# still records ('chunkdata.bin_-1', 'chunkdata.bin', -1) via str2num,
-# and Python's isdigit-based parse answers ('', None) -- both languages
-# refuse the name, only the layer differs. See DID-python#69.
-DEVIATIONS: dict[str, tuple[str, int | None, str]] = {
-    "chunkdata.bin_-1": (
-        "",
-        None,
-        (
-            "MATLAB's str2num parses a leading minus; Python's isdigit() does not. "
-            "Both refuse the name, at different layers. Removed here once "
-            "VH-Lab/DID-matlab#198 lands and MATLAB's parse becomes strict too."
-        ),
-    ),
-}
+# Empty as of DID-matlab#198 / DID-python#69: MATLAB tightened its parse
+# from str2num (which EVALUATES) to a strict all-digits check, matching
+# Python's isdigit-based parse. The '_-1' vector that used to be a
+# MATLAB-only member is now refused at the parse in both languages.
+DEVIATIONS: dict[str, tuple[str, int | None, str]] = {}
 
 
 def _recorded_index(case):