feat(chunks): rilevamento universale has_math/has_table + dollarmath plugin

- parser.py: integra dollarmath_plugin (graceful fallback) — blocchi $$…$$ → token math_block atomico
- segmenter.py: gestisce math_block → Block(kind=math, atomic=True, plain_text=[formula])
- packer.py: has_table rileva anche <table> in blocchi html; has_math rileva $$ e \begin{ nel contenuto
- +11 test (parser×3, segmenter×3, packer×5) — totale 65 test

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

chunks/parser.py

@@ -2,6 +2,12 @@ from __future__ import annotations
 from markdown_it import MarkdownIt
 from markdown_it.token import Token
 
+try:
+    from mdit_py_plugins.dollarmath import dollarmath_plugin as _dollarmath
+    _HAS_DOLLARMATH = True
+except ImportError:
+    _HAS_DOLLARMATH = False
+
 
 def parse(source: str) -> tuple[list[Token], list[str]]:
     """Parsa Markdown in token stream con source map.
@@ -11,6 +17,8 @@ def parse(source: str) -> tuple[list[Token], list[str]]:
         lines:  righe sorgente (0-indexed) per ricostruzione testo esatto
     """
     md = MarkdownIt().enable("table")
+    if _HAS_DOLLARMATH:
+        md = md.use(_dollarmath, allow_labels=False, allow_space=False)
     tokens = md.parse(source)
     lines = source.splitlines()
     return tokens, lines