Test floating point edge cases

Writing from ASCII into netCDF doubles and reading doubles back out into Java results in some floating point issues when looking up the lon/lat indices, in that the returned index does not always correspond to the first value less than or equal to the target value as one might expect. For example, looking up a target value that should be equal to a value in the list should return the index of that value, but thanks to floating point dust, the index may be returned for the next lower value instead.