The goal is to help writing methods by associating a methods file to
some worflow outputs. Methods from one rule may copy methods from
upstream before appending their part.

Also removed id to name conversion from htseq-counts wrapper.

Also made one main requirements.txt file.

One filters unique reads, the other filters reads in a size range.

Also fixed a bug.

Normal python variables are actually not visible in the wrapper.

It seems that shell.prefix do not stack. Only the last one is used.

This currently assumes that the cluster nodes all have names starting
with "tars". Not tested.

Also changed the way to cleanup temporary dir when making bigwig files.

Also removed some debugging output that was written in a unique file in
/tmp, which prevented the small RNA annotation to be performed by a new
user without erasing this file.

Factorized linking to raw data.

It turns out that using `include:` can fail when there are symlinks in
the path to the snakefile.