Hi Guix,

attached is a rewritten CRAN importer (and updater) with additional
test.

I only recently found out that the alternative to parsing the HTML
description page for an R package does not require downloading and
extracting the DESCRIPTION file from the tarball — the DESCRIPTION file
is hosted as plain text on CRAN.

So I rewrote the CRAN importer to do this:

  * download the DESCRIPTION file for a given package
  * break it up into a simple alist
  * transform the alist into a package expression

This is much simpler than the sxml hackery we did before and the code
can be reused to write an importer for Bioconductor, a popular,
versioned R package repository for bioinformatics packages.[1]

Along the way I made the importer more robust and added a few cosmetic
improvements (such as adding a properties field containing the upstream
package name as suggested by Ludo).  I already used the new importer to
process *all* packages from CRAN.  It didn’t produce any errors
(although post-processing of the resulting expressions is inevitable).

Once this patch has been reviewed and accepted I’ll write an importer
for Bioconductor.

~~ Ricardo