Raw Input		Desired Output
line 1 line 2 line 3 line 4 line 5 line 6		line 1 line 2 line 3 line 4 line 5 line 6

Script and Comments

Script₁

[ 1] :loop0
[ 2] $!N
[ 3] $!b loop0
[ 4] :loop1
[ 5] s/\n([^\n]*\n)/ \1/g
[ 6] t loop1

Comments

The `-r' option of GNU sed must be used or we have to escape the parentheses used in Step [5].
Steps [1] thru [3] constitute a loop which iterates until all lines have been read to Pattern Space.

Steps [4] thru [6] constitute a loop which deletes all except the last newline characters, let's take the sample as an example:

After n-th iteration	contents
Initial	line 1 \n line 2 \n line 3 \n line 4 \n line 5 \n line 6
1	line 1 line 2 \n line 3 line 4 \n line 5 \n line 6
2	line 1 line 2 line 3 line 4 \n line 5 \n line 6
3	line 1 line 2 line 3 line 4 line 5 \n line 6