Extract rule: template-modifier-name-case

[NullVoxPopuli]

Split from #2371.

[johanrd]

Review from claude: comparing against ember-template-lint modifier-name-case

dasherize is buggy for leading-capital and consecutive-capital names. The PR's implementation blindly prepends - before every uppercase letter:

str.replaceAll(/([A-Z])/g, '-$1').toLowerCase()

The original has guards for index 0 and non-alpha preceding chars:

// index === 0 or previous char is not a letter → just lowercase, no dash
if (index === 0 || !ALPHA.test(key[index - 1])) {
  return char.toLowerCase();
}
return `-${char.toLowerCase()}`;

Results:

Input	Original	PR
FooBar	foo-bar	-foo-bar
XFoo	x-foo	-x-foo
HTMLParser	htmlparser	-h-t-m-l-parser

The current test cases all use lowercase-initial camelCase (didInsert, fooBar) so the bug doesn't surface. It would break any PascalCase or acronym-prefixed modifier name. The original also handles :: →
/ for namespaced paths (e.g. Foo::BarBaz → foo/bar-baz).

Fix: port the original's dasherize-component-name helper, or at minimum add the index-0 and consecutive-capital guards.

Unused broken messageId. meta.messages.dasherized uses {{dasherizeModifierName}} for both the before and after placeholders — it would render the same value twice. It's never actually used (the rule
builds messages via generateErrorMessage inline). Either fix the template to use distinct placeholders and switch to messageId, or remove it.