Fix half_impl::float_to_half_rtne(float) warning: '<<' causes overflow

Fixed Visual Studio 2019 Code Analysis (C++ Core Guidelines) warning
C26450 from inside `half_impl::float_to_half_rtne(float)`:
> Arithmetic overflow: '<<' operation causes overflow at compile time.

Eigen/src/Core/arch/Default/Half.h

diff --git a/Eigen/src/Core/arch/Default/Half.h b/Eigen/src/Core/arch/Default/Half.h
index bbc15d4..60f1974 100644
--- a/Eigen/src/Core/arch/Default/Half.h
+++ b/Eigen/src/Core/arch/Default/Half.h
@@ -472,7 +472,9 @@
       unsigned int mant_odd = (f.u >> 13) & 1; // resulting mantissa is odd
 
       // update exponent, rounding bias part 1
-      f.u += ((unsigned int)(15 - 127) << 23) + 0xfff;
+      // Equivalent to `f.u += ((unsigned int)(15 - 127) << 23) + 0xfff`, but
+      // without arithmetic overflow.
+      f.u += 0xc8000fffU;
       // rounding bias part 2
       f.u += mant_odd;
       // take the bits!