Enhancement

Our current implementation of cast(xx as Decimal) will check overflow for each input, actually, if source is a acurate type, we can use meta data to check if this cast can overflow, for example,
cast(int64 to Decimal(20,0))
Since the length of Int64.max_value/Int64.min_value is 20, when casting Int64 to Decimal(20,0), it should never overflow, so there is no need to check overflow for every input.