mysql delete row "almost" duplicated

for a SELECT

SELECT * FROM registros
  WHERE id 
  IN (
   SELECT t.id FROM (
     SELECT 
       DATE_FORMAT(r.fecha,'%Y-%m-%d %H:%i:00') AS fecha_sin_segundos,
       MIN(r.id) AS id 
     FROM registros r 
     GROUP BY fecha_sin_segundos
    ) AS t
);

for a DELETE (eye make backup) we put a NOT IN

DELETE FROM registros
  WHERE id 
  NOT IN (
   SELECT t.id FROM (
     SELECT 
       DATE_FORMAT(r.fecha,'%Y-%m-%d %H:%i:00') AS fecha_sin_segundos,
       MIN(r.id) AS id 
     FROM registros r 
     GROUP BY fecha_sin_segundos
    ) AS t
);

Take the record with id less, if you want to keep / show the id greater replace the MIN(r.id) by MAX(r.id)

to make a backup and test the cleaning you can use the following:

CREATE TABLE registros_backup LIKE registros;
INSERT registros_backup SELECT * FROM registros;

and you work on registros_backup

edited: For tables with many records you may want to create a new table and copy the data clean (add the "backup" with the select):

CREATE TABLE registros_limpio LIKE registros;
INSERT registros_limpio 
SELECT * FROM registros
  WHERE id 
  IN (
   SELECT t.id FROM (
     SELECT 
       DATE_FORMAT(r.fecha,'%Y-%m-%d %H:%i:00') AS fecha_sin_segundos,
       MIN(r.id) AS id 
     FROM registros r 
     GROUP BY fecha_sin_segundos
    ) AS t
);

You can see the pattern that each record is repeated once, then eliminate the even ids. So you would only have one record left, always BCK (BACKUP) of your table.

delete from prueba WHERE mod(id,2) = 0

Best regards,