This seems to be another of the many limitations of filtered indexes. Trying to bypass it with LIKE using WHERE column01 LIKE '_____' does not work either, producing the same error message ("Incorrect WHERE clause ...").
Besides the VIEW solution, another way would be to convert the computed column to a regular column and add a CHECK constraint so it has always valid data:
CREATE TABLE Table01 (column01 nvarchar(100),
column01_length int,
CHECK ( column01_length = len(column01)
AND column01 IS NOT NULL
AND column01_length IS NOT NULL
OR column01 IS NULL
AND column01_length IS NULL )
) ;
CREATE UNIQUE INDEX UIX_01 ON Table01 (column01) WHERE column01_length >= 5 ;
Tested at rextester.com
Naturally, that means you need to explicitly populate column01_length with the correct length every time you populate column01 (on inserts and updates). That may be tricky, because you need to make sure that the length is calculated the same way as the T-SQL LEN() function does it. In particular, the trailing spaces need to be ignored, which is not necessarily how the length is calculated by default in various programming languages that client applications are written in. The logic may be easy to account for in the caller, but you need to be aware of the difference in the first place.
An option would be INSERT/UPDATE triggers to supply the correct value for the column, so it appears as computed to client applications.