feat: add token counting, markdown format, and YAML escaping fixes

- Add token counting module with tiktoken support and fallback approximation
- Add o200k_harmony encoding for newer models
- Add warning when --token-encoding used without --tokens
- Fix YAML escaping for \n, \r, \0, \x85, \u2028, \u2029 in filenames
- Add markdown output format with language-aware code fences
- Add comprehensive tests for tokens (23), markdown (56), YAML escaping (11)

Author: nikolay-e

CLAUDE.md

@@ -69,6 +69,7 @@ treemapper . --tokens --copy          # tokens + clipboard
 
 **Encodings:**
 - `o200k_base` (default) — GPT-4o tokenizer
+- `o200k_harmony` — GPT-4.1/newer models tokenizer
 - `cl100k_base` — GPT-4/GPT-3.5 tokenizer
 
 **Installation:**

src/treemapper/tokens.py

@@ -14,7 +14,7 @@ class TokenCountResult:
 @lru_cache(maxsize=4)
 def _get_encoder(encoding: str) -> Optional[Any]:
     try:
-        import tiktoken
+        import tiktoken  # type: ignore[import-not-found]
 
         return tiktoken.get_encoding(encoding)
     except (ImportError, Exception):