f2fs: lookup neighbor extent nodes for merging later
In __lookup_extent_tree_ret we will not try to find neighbor nodes if
we find the target node, in this condition, we will lost the chance to
merge the new mapping with exist extent node later.
So our extent cache of inode will be fragmented after overwrite exist
file, we can see the number of extent node increases intensively in
following test case: