Fix UInt128 to double conversion for values in [2^88, 2^104)

Change threshold check from (value._upper >> 24) to (value._upper >> 40)
to correctly handle values up to 2^104 in the intermediate precision branch.

Author: sdcb

src/libraries/System.Private.CoreLib/src/System/UInt128.cs

@@ -275,7 +275,7 @@ public static explicit operator double(UInt128 value)
                 // for the precision loss that double will have. As such, the lower value effectively drops the
                 // lowest 24 bits and then or's them back to ensure rounding stays correct.
 
-                double lower = BitConverter.UInt64BitsToDouble(TwoPow76Bits | ((ulong)(value >> 12) >> 12) | ((value._lower & 0xFFFFFF) != 0 ? 1UL : 0UL)) - TwoPow76;
+                double lower = BitConverter.UInt64BitsToDouble(TwoPow76Bits | ((ulong)(value >> 12) >> 12) | (value._lower & 0xFFFFFF)) - TwoPow76;
                 double upper = BitConverter.UInt64BitsToDouble(TwoPow128Bits | (value._upper >> 12)) - TwoPow128;
 
                 return lower + upper;