The CSV file is malformed. There is a column where there is a
valid value a some of the rows but with no column header, its blank. Also, there a dozen of so blank columns at the end of each row. These issues seem to mess up the parser. Fixing these issues, the file imports fine in both Connections 8.1.7 and the upcoming 8.2 version. If you are using Excel to create the CSV file, I suggest using LibreOffice as Excel can create some nasty CSV files.
Not sure why this would cause duplicate category creation … my best guess is that the empty columns were somehow being appended creating a “unique” name which when checked for did not exist so a new one would be created.
Probably the quickest way to cleanup is to go to the Categories admin page click the select all checkbox and then uncheck the categories to keep and then bulk delete the “duplicates”.