It appears to me as the easiest, most straightforward solution, but please correct me if I'm wrong.
Instead of having a fulltext index on all individual columns of a table, isn't it better to just generate one single wide computed column and run the fulltext index against that only?
It appears to me that it gets rid of all the issues of having multiple columns, incl. that I can't search "x AND y" as this will not match a row with "x" present in column 1 and "y" present in column 2.
Any counterarguments?