codecaliper v0.2.0

• package 0.2.0 · spec 1.2.0 · grammars: tree-sitter-python 0.25.0,
  tree-sitter-java 0.23.5, tree-sitter-go 0.25.0 (binding tree-sitter 0.26.0)
• The package MINOR bump reflects a third language and a new runtime dependency
  (tree-sitter-go); no previously emitted number changed, so the spec bump is a
  MINOR too (1.1.0 -> 1.2.0).

Go language support (spec 1.2.0)

• Go is the third language, and the first added since the extension seam was
  tightened. It is confined to the languages/ package: languages/go.py plus
  its registry entry, no edit to the metric engines, readability, model, API,
  CLI or grammar loader. Nine new rulings (CC-GO-0001..0005, COG-GO-0001,
  TOK-GO-0001, BW-GO-0001, BW-GO-0002); every other rule is a reused -ALL-
  ruling. Spec bumped 1.1.0 -> 1.2.0 (MINOR: additive, no existing corpus value
  changed).
• Notable Go-specific rulings: for is the only loop keyword and covers every
  loop form; select communication cases are a distinct cyclomatic ruling from
  switch cases; labeled break/continue and goto are a fundamental cognitive
  +1 in both modes; true/false/nil/iota are predeclared IDENTIFIERS, not
  keywords (a spec-faithful divergence from Java's BW keyword set).
• Cross-checked against lizard (the Go cyclomatic oracle), which concurs on
  every Go probe. Eight hand-computed corpus cases. docs/adding-a-language.md
  documents the whole procedure with Go as the worked example.

A new language is now confined to the languages/ package

• No emitted number changed: this is a plumbing refactor, and the spec-drift
  gate proves every corpus value is untouched.
• Two duplications of the language set that used to sit outside languages/
  were folded back in. syntax/grammars.py no longer keeps a _GRAMMAR_MODULES
  dict; load(language_name, module) takes the tree-sitter package name from
  the adapter, which now carries it as LanguageAdapter.grammar_module. cli.py
  no longer hardcodes --lang choices; it derives them from the registry via
  available_languages(). Adding a language therefore touches only a new adapter
  module and the registry in languages/__init__.py; the CLI and the grammar
  loader need no edit. (Internal API: grammars.load gained a required module
  argument.)