feat: air-gapped deployment guide, no implicit downloads

- Ship tiktoken's cl100k_base inside the package and build the encoding
  from it, so token counting never downloads anything.
- Default EMBEDDINGS_CACHE_DIR to <data home>/models instead of FastEmbed's
  temp dir, and read tokenizer.json and repo metadata from that cache, so
  a model downloads once and survives reboots.
- TTS_PROVIDER=none and STT_PROVIDER=none switch the speech features off:
  the endpoints return 404, audio files fail to ingest with a clear
  message, /api/config reports tts_available/stt_available, and the UI
  hides the Speak and microphone buttons.
- Drop the Google Fonts Roboto import from the web UI.
- prefetch-models fills the cache the app reads; verify-offline checks the
  packaged encoding.
- Docs: new Air-Gapped Deployment guide, settings and cache notes.

tests/test_utils.py

@@ -40,6 +40,47 @@ class TestGetEncoding:
         enc2 = get_encoding()
         assert enc1 is enc2
 
+    @pytest.mark.unit
+    def test_loads_the_packaged_encoding_without_network(self, monkeypatch):
+        """Token counting runs on every chat; it must never download the encoding."""
+        import tiktoken
+        import tiktoken.load
+
+        from docsgpt import utils
+
+        def no_download(*args, **kwargs):
+            raise AssertionError("tiktoken tried to fetch cl100k_base")
+
+        monkeypatch.setattr(utils, "_encoding", None)
+        monkeypatch.setattr(tiktoken.load, "read_file", no_download)
+        monkeypatch.setattr(tiktoken, "get_encoding", no_download)
+        assert utils.get_encoding().encode("hello world") == [15339, 1917]
+
+    @pytest.mark.unit
+    def test_packaged_encoding_matches_the_tiktoken_definition(self, monkeypatch):
+        from tiktoken_ext import openai_public
+
+        from docsgpt import utils
+
+        monkeypatch.setattr(utils, "_encoding", None)
+        monkeypatch.setattr(openai_public, "load_tiktoken_bpe", lambda *args, **kwargs: {})
+        reference = openai_public.cl100k_base()
+        encoding = utils.get_encoding()
+        assert encoding.name == reference["name"]
+        assert encoding._pat_str == reference["pat_str"]
+        assert encoding._special_tokens == reference["special_tokens"]
+
+    @pytest.mark.unit
+    def test_corrupt_packaged_encoding_is_rejected(self, monkeypatch, tmp_path):
+        from docsgpt import utils
+
+        corrupt = tmp_path / "cl100k_base.tiktoken"
+        corrupt.write_bytes(b"IQ== 0\n")
+        monkeypatch.setattr(utils, "_encoding", None)
+        monkeypatch.setattr(utils, "_CL100K_BASE_FILE", corrupt)
+        with pytest.raises(ValueError, match="cl100k_base"):
+            utils.get_encoding()
+
 
 class TestGetGptModel: