- Built-in functions now handle unit values more safely, with clearer errors when a function cannot use a given unit.
- Fixed modulo so values with physical units fail clearly instead of producing incorrect results.
