fix: filter out SESSION START / SESSION END marks in parse_structured_bin, and also add status feedback.

parsers/s3_analyzer.py

@@ -344,9 +344,13 @@ def parse_structured_bin(bin_blob: bytes) -> list[MarkSplit]:
             s3_bytes += length
         elif rec_type == _REC_MARK:
             label = payload.decode("utf-8", errors="replace")
-            marks.append(MarkSplit(label=label,
-                                   bw_byte_offset=bw_bytes,
-                                   s3_byte_offset=s3_bytes))
+            # Skip auto-generated bridge lifecycle marks — only keep user marks
+            if label.startswith("SESSION START") or label.startswith("SESSION END"):
+                pass
+            else:
+                marks.append(MarkSplit(label=label,
+                                       bw_byte_offset=bw_bytes,
+                                       s3_byte_offset=s3_bytes))
 
         pos = payload_end