The core doco perlpragma explains how to do this.

The main drawback to that approach for my little project is the additional overhead of having to turn perl scalars into Math::Ryu objects and then overloading their stringification.
It's probably not such a big deal, but then neither is the original issue ;-)
BTW, in the Math::Ryu github repo, the function is named "pany", not "palt" - neither of which have yet appeared in a cpan release of the module.

AFAICT, the "integer" pragma doesn't rely on module objects and operator overloading.
I speculate that it's functionality is built into the perl source, thereby avoiding the need for objectification and operator overloading.

OTOH, "bignum", "bigrat", "bigint" and "bigfloat" are clearly built on Math::BigInt, Math::BigRat and Math::BigFloat, and use the overloading that those 3 modules provide.

It's all interesting stuff ...

Cheers,
Rob