Fix float16 inequality comparing raw bits as an integer

[taking-lying-flat]

float16::operator!= currently passes rhs.x to operator==. This selects the integer overload and converts the right operand's raw bit pattern as a numeric integer, instead of comparing the two float16 values.

For example:

const gloo::float16 one(1);
const gloo::float16 numericBits(15360); // 15360 is the raw bit pattern of half(1).

one == one;         // true
one != one;         // currently true; should be false
numericBits == one; // false
numericBits != one; // currently false; should be true

Pass rhs itself to the existing equality operator. This one-line change makes inequality its logical complement and preserves the current equality semantics, including its treatment of special bit patterns.

Validation against the actual modified header, compiled with GCC 15.2.0, C++20 and -O2:

• Exhaustively checked all 65,536 bit patterns for (a != a) == !(a == a): 65,533 inconsistencies before, zero after.
• Checked 100,000 deterministic pseudorandom pairs for inequality/equality complementarity and inequality symmetry: all passed after the fix.
• Both concrete examples above now return the expected results.
• clang-format 18.1.8 --dry-run --Werror gloo/types.h and git diff --check passed.

Only this production line is changed; no test files are included. The comparison checks ran on CPU; the full repository suite and CUDA build were not run.

[taking-lying-flat]

Consolidated into #527 to keep these small Gloo correctness fixes in one review. The same one-line float16 inequality fix is included there as a separate commit, with the comparison checks rerun on the combined version. Closing this duplicate PR.