refactor(schema): align array overload structure

[SrHenry]

Summary

Merges the split ValidatorMap<T> and TypeGuard<T> | StandardSchemaV1<T, T> overloads in _fn and OptionalizedArray into a single combined overload, matching the structure already used in ArraySchema.

Changes

• src/validators/schema/array.ts: merged _fn overloads (2 → 1), merged OptionalizedArray overloads (2 → 1), reordered (tree: {}) last
• src/validators/schema/types/ArraySchema.ts: reordered (tree: {}) last (already done in fix(schema): preserve array element inference #46, included here for completeness)
• src/validators/__tests__/array.spec.ts: type inference + runtime tests (13 tests)

Verification

• All 377 tests pass
• Typecheck clean
• No circular dependencies
• Lint/format clean

Related

• Closes refactor(schema): align _fn/OptionalizedArray overload structure with ArraySchema #47
• fix(types): Infer drops | undefined from FluentOptionalSchema #48 was closed as not-a-bug (Infer correctly preserves | undefined) — test updated accordingly

[SrHenry]

Review notes (low severity)

1. Merged overload — structurally ambiguous inputs: Merging ValidatorMap<T> | TypeGuard<T> | StandardSchemaV1<T, T> into a single overload means TypeScript infers T from a union type rather than independently per overload. If a value is simultaneously callable and has a ValidatorMap-like shape (e.g., a callable object with validator properties), the merged form could infer a wider T than the split overloads would. In practice, TypeGuard (function) and ValidatorMap (plain object record) are structurally disjoint, so this is unlikely to matter.

2. No test for callable-object edge case: A test passing a callable ValidatorMap-like object would explicitly confirm the merge does not change inference behavior for ambiguous inputs. Risk is low given the structural disjointness, but worth noting for completeness.