fix(parser,vectorstore): stop the first-request downloads in a warmed install

Three things still reached the network from a container whose models were
baked in:

- tiktoken fetched cl100k_base from openaipublic.blob.core.windows.net on
  every fresh container (its cache defaulted to /tmp), and token accounting
  calls it on every chat. prefetch_models now warms it too; the image sets
  TIKTOKEN_CACHE_DIR.
- The chunker loaded its tokenizer with Tokenizer.from_pretrained, which
  revalidates the revision with a HEAD request per process start and stalls
  for the etag timeout (10 s) when huggingface.co is unreachable. It now reads
  tokenizer.json from the hub cache first and only downloads on a miss; the
  repo-metadata read for models outside the registry does the same.
- tldextract fetched the public suffix list on the first web crawl; the
  bundled snapshot is used instead.

application/scripts/verify_offline.py exercises these paths (and docling's
conversion when the extra is installed) so an image can be checked with
docker run --network none.

tests/scripts/test_prefetch_models.py

@@ -94,3 +94,18 @@ class TestMain:
         with patch.object(prefetch_models, "prefetch", return_value=[]) as spy:
             prefetch_models.main(["granite-97m"])
         assert spy.call_args.args[1] == "/app/models"
+
+
+class TestPrefetchTiktoken:
+    def test_warms_every_listed_encoding(self):
+        """The image sets TIKTOKEN_CACHE_DIR; warming fills it at build time."""
+        fake = MagicMock()
+        module = types.ModuleType("tiktoken")
+        module.get_encoding = fake
+        with patch.dict(sys.modules, {"tiktoken": module}):
+            fetched = prefetch_models.prefetch_tiktoken()
+        assert fetched == list(prefetch_models.TIKTOKEN_ENCODINGS)
+        assert [c.args[0] for c in fake.call_args_list] == list(prefetch_models.TIKTOKEN_ENCODINGS)
+
+    def test_cl100k_is_the_encoding_token_counting_uses(self):
+        assert "cl100k_base" in prefetch_models.TIKTOKEN_ENCODINGS