Here is the logic of the output:
Files with a common first number (first level grouping) are identical — they contain the same data.
Files with a common first and second number (second level grouping) are physically the same file — writing to one file will affect all.
The third number is there only to enumerate members of the second level group, and is omitted when there is only one member.
For human readability, each first level group is preceded by a comment of the form:
-- #<first level group number> -- <filesize> --