make pypi html parser regex less greedy

[jbouzekri]

This makes the regex to find the download link of a python package less greedy.

Before, on this string :

<a href="https://files.pythonhosted.org/packages/ba/bb/dfa0141a32d773c47e4dede1a617c59a23b74dd302e449cf85413fc96bc4/requests-0.2.0.tar.gz#sha256=813202ace4d9301a3c00740c700e012fb9f3f8c73ddcfe02ab558a8df6f175fd">requests-0.2.0.tar.gz</a><br/>
<a href="https://files.pythonhosted.org/packages/4b/ad/d536b2e572e843fda13e4458c67f937b05ce359722c1e4cdad35ba05b6e3/requests-0.2.1.tar.gz#sha256=d54eb33499f018fc6bd297613bf866f8d134629c8e02964aab6ef951f460e41e">requests-0.2.1.tar.gz</a><br/>

for the package requests-0.2.1.tar.gz it would have matched the download link of package requests-0.2.0.tar.gz which is not intended behavior (in fact it always download the first link found in the page)