Fix a bug that exponent overflow is ignored in add, sub, mult and div operation

[tompng]

Fixes add/sub overflow bug

x # => 0.1e9223372036854775793
x+x       #=> 0.0  (Expected: huge positive value or infinity)
x-(-x)    #=> 0.0  (Expected: huge positive value or infinity)
(-x)+(-x) #=> -0.0 (Expected: huge negative value or -infinity)
(-x)-x    #=> -0.0 (Expected: huge negative value or -infinity)

Fixes div and mult overflow/underflow bug: wrong sign

large_x #=> 0.1e4611686018427387905
small_x #=> 0.1e-9223372036854775807

large_x * (-large_x) #=> Infinity (Expected: -Infinity)
small_x * (-small_x) #=> 0.0 (Expected: -0.0)
large_x.div(-small_x, 10) #=> Infinity (Expected: -Infinity)
small_x.div(-large_x, 10) #=> 0.0 (Expected: -0.0)

No test for add/sub because it is hard to make a reproducible bigdecimal with large exponent, some unrelated bug is a blocker of writing test.