Fix sign-extension in FNV-1/FNV-1a byte processing; add std::byte support

[Copilot]

update() was casting range elements directly to std::size_t, causing sign-extension on platforms where char is signed (bytes ≥ 0x80 would expand to e.g. 0xFFFFFFFFFFFFFF80), producing incorrect and platform-dependent hashes.

Changes

• fnv_1.h / fnv_1a.h: update() now uses if constexpr to dispatch byte normalization:
  • std::byte → std::to_integer<std::uint8_t>() (idiomatic)
  • all other types → static_cast<std::uint8_t>() (strips sign-extension)
  • Added <cstdint> and <type_traits> includes

for (auto const &byte : data) {
    using value_type = std::remove_cvref_t<decltype(byte)>;
    std::uint8_t b;
    if constexpr (std::is_same_v<value_type, std::byte>) {
        b = std::to_integer<std::uint8_t>(byte);
    } else {
        b = static_cast<std::uint8_t>(byte);
    }
    hash_ ^= b;
    hash_ *= prime();
}

• fnv_1_tests.cpp / fnv_1a_tests.cpp: Added high_byte_values_test (std::vector<signed char> with values ≥ 0x80, e.g. static_cast<signed char>(0x80) = -128) and std_byte_test (std::vector<std::byte>) — both asserting identical results against known-correct 32-bit and 64-bit reference values. Using signed char directly exercises the sign-extension path the fix guards against.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[rwindegger]

@copilot is this pr ready?

[Copilot]

@copilot is this pr ready?

Yes — all review threads are resolved and all CI checks on this branch are green. The PR is ready to merge.