- All places these are used already handle magnitude checks correctly
- As a "bit-twiddling" method, low_bits() truncates
- As a "semantics preserving type conversion" method, to_uint64() asserts
- Add low_32_bits() as a "bit-twiddling" non-asserting to_integer() equivalent
The issue was that redeclaring the hash template in the code somehow seems to shadow the declaration that is done in <functional> and <memory> and afterwards the template specialization of the hash function seems to use our redefined template instead of libc++’s one (most likely due to the fact that libc++ uses an inline namespace inside std)
