This could allow users on 32-bit platforms to perform mathematical operations with native integers and Int64 types. If the result of an operation fits in the range of a 32-bit integers, it would probably make sense to return that instead of an Int64 object (like BSON decoding); however, we should see if there may be edge cases involved.

Alternatively, we may decide to simply not implement this and suggest that users resort to GMP for computations (akin to using BCMath for Decimal128).