extract: auto-detect gzip by magic bytes, not just .gz extension

Files compressed with gzip but named without .gz (e.g. activity.gpx
containing gzip data) now decompress transparently.

bincio/extract/parsers/base.py

@@ -27,8 +27,14 @@ class BaseParser(ABC):
 
         raw_bytes is the original file content (used for hashing).
         decompressed_bytes is what parsers should actually parse.
+
+        Gzip is handled both by extension (.gz) and by magic bytes (0x1f 0x8b),
+        so files that are gzip-compressed but named without .gz still parse correctly.
         """
         raw = path.read_bytes()
-        if path.suffix == ".gz":
-            return raw, gzip.decompress(raw)
+        if path.suffix == ".gz" or raw[:2] == b'\x1f\x8b':
+            try:
+                return raw, gzip.decompress(raw)
+            except Exception:
+                pass  # not actually gzip despite the magic bytes — fall through
         return raw, raw