Numeric CSV fields should throw exceptions with invalid data

Currently, a NumberFormatException when coercing CSV data into a numeric type will result in a null value. This is to catch the case where the empty string is passed and should be considered null rather than a value.

The unintended consequence is that any invalid string will result in a null value. If the field type is Long and a decimal value is present, this will also result in a null that could be replaced by the default value for the field. That would mean that the import process would silently drop data that appears valid and replace it with the default. Instead, this case should throw an exception.