Remove Type::some() — an identity function masquerading as a type constructor

[MidnightDesign]

Problem

Type::some() was a public static factory that returned its argument unchanged:

public static function some(self $some): self
{
    return $some;
}

Sitting next to option() (a real Option<T> wrapper) and none() (a real None type), it read as "the Some branch of an Option" but constructed no Some type at all. Type::some(Type::int()) was exactly Type::int(), so calls like x.isSubtypeOf(Type::some(...)) behaved nothing like they read — a silent footgun. Some appears nowhere in the README's user-facing type list.

Change

• Removed Type::some() from the public surface.
• Removed its only test. The equalsCases provider held a single case, 'Some<T> == T', whose whole purpose was to assert the identity behavior being removed; with it gone the provider would be empty, so the provider and its testEquals consumer go together. equals()'s positive path stays covered by testAliasTypeEqualsAliasTarget, and the Some<...> type-parsing path (notEqualsCases) is untouched.

Deliberately left alone

TypeConstructor::Some and the parser's resolveSome() are legitimate — they back the Some<T> type syntax that Option<T> builds on. This PR targets only the misleading public factory.

BC impact

Structural break (public method removal). Scoped for the 0.3.0 breaking release.

Verification

• PHPUnit: 1130 tests OK
• Psalm / PHPStan: no errors · PHP-CS-Fixer: clean · composer check-deps: no unknown symbols
• Infection on the diff (vs master, 100% MSI): deletion-only change, no mutants generated — passes under --ignore-msi-with-no-mutations

Closes #93

🤖 Generated with Claude Code