I can only suppose that these are the problems, which are usually asked at an interview for a job. As far as I can guess, this question would be answered by Joe Celko in the following way: error of design is quite evident, as one attribute is split into two.But let us put aside the issues of designing and further consider the solution of such problem.
Example of results after importing data from KML file (in English, but other language options are available): You can download an example file at: Example KML file.
Let's create a test table and populate it with some data: I know of three methods of solution employing standard means of interactive SQL language.
The solution is quite evident and not calling for any comments.
For the first two solutions (with use of UNION and FULL JOIN) the scanning procedure is carried out twice, while for the last one (UNPIVOT) only once, that accounts for double advantage in terms of performance.
The Cartesian product of the T table with derived table of 2 rows gives "doubling" (each row of T is presented 2 times, for =1 and =2).