nixguix: Improve further tarball detection

The current content type detection was a bit off mostly for content which includes
charset. This commit fixes it.

Related to T3781

swh/lister/nixguix/lister.py

@@ -121,7 +121,7 @@ PageResult = Tuple[ArtifactType, Union[Artifact, VCS]]
 VCS_SUPPORTED = ("git", "svn", "hg")
 
 # Rough approximation of what we can find of mimetypes for tarballs "out there"
-POSSIBLE_TARBALL_MIMETYPES = set(MIMETYPE_TO_ARCHIVE_FORMAT.keys())
+POSSIBLE_TARBALL_MIMETYPES = tuple(MIMETYPE_TO_ARCHIVE_FORMAT.keys())
 
 
 def is_tarball(urls: List[str], request: Optional[Any] = None) -> Tuple[bool, str]:
@@ -218,7 +218,7 @@ def is_tarball(urls: List[str], request: Optional[Any] = None) -> Tuple[bool, st
             logger.debug("Content-Type: %s", content_type)
             if content_type == "application/json":
                 return False, urls[0]
-            return content_type in POSSIBLE_TARBALL_MIMETYPES, urls[0]
+            return content_type.startswith(POSSIBLE_TARBALL_MIMETYPES), urls[0]
 
         raise ArtifactNatureUndetected(
             f"Cannot determine artifact type from url <{url}>"

swh/lister/nixguix/tests/data/nixpkgs-swh_sources.json

@@ -79,6 +79,13 @@
       "svn_url": "https://code.call-cc.org/svn/chicken-eggs/release/5/iset/tags/2.2",
       "svn_revision": 39057
     },
+    {
+      "type": "url",
+      "urls": [
+        "http://git.marmaro.de/?p=mmh;a=snapshot;h=431604647f89d5aac7b199a7883e98e56e4ccf9e;sf=tgz"
+      ],
+      "integrity": "sha256-G/7oY5qdCSJ59VlwHtIbvMdT6+mriXhMqQIHNx65J+E="
+    },
     {
       "type": "url",
       "urls": ["svn://svn.code.sf.net/p/acme-crossass/code-0/trunk"],

swh/lister/nixguix/tests/test_lister.py

@@ -173,6 +173,10 @@ def test_lister_nixguix_ok(datadir, swh_scheduler, requests_mock):
         "https://api.github.com/repos/trie/trie",
         [{"json": {"html_url": "https://github.com/trie/trie.git"}}],
     )
+    requests_mock.head(
+        "http://git.marmaro.de/?p=mmh;a=snapshot;h=431604647f89d5aac7b199a7883e98e56e4ccf9e;sf=tgz",
+        headers={"Content-Type": "application/gzip; charset=ISO-8859-1"},
+    )
 
     expected_visit_types = defaultdict(int)
     # origin upstream is added as origin