BigDecimal-RS round function cannot handle values with high precision

[ekump]

While attempting to implement the natural logarithm function I quickly ran into an issue where the round function in BigDecimal-RS panics. This is because it attempts to cast the BigInt component of the BigDecimal to an i128, which has a max value of 170,141,183,460,469,231,731,687,303,715,884,105,727. Before we can move forward with the natural logarithm functionality we are going to have to implement rounding in BigDecimalMath that isn't bound by i128 maximum values.