Add Integer32 and UInteger32 for fixed-width arithmetic

Add optional Integer32 and UInteger32 value types for signed and unsigned 32-bit arithmetic. These make it easier to port algorithms that rely on overflow at exactly 32 bits, including checksums, hashes, pseudorandom sequences, and compatibility calculations for existing data formats.

Arithmetic on these types should preserve the selected width and wrap consistently, allowing ordinary expressions instead of a conversion helper around every intermediate result. Both types should provide parsing, formatting, conversions, and bitwise operations.

Keep the existing 64-bit Integer as the default general-purpose integer. Existing projects and APIs should retain their current behaviour; developers only need the new types when they explicitly choose them.