mesa: when INT_MAX and INT_MIN get converted into float, precision lowers
Be explicit about the conversion values.
Fixes clang warning:
../src/mesa/main/get.c:86:31: warning: implicit conversion from 'int' to 'float' changes value from 2147483647 to 2147483648 [-Wimplicit-const-int-float-conversion]
Not 100% sure it's the right solution, but seems it could work. Anyway, this should be fixed this way or another, as the code doesn't represent the right math.