Decouple the download process

[Jordi Mas]

Limitation: Current the process of downloading, converting and building the translation memories is coupled

Solution: We need to split it on the 3 separate process

• Create separate modules and classes for the 3 process
• Most of the code for this currently lives in fileset.py

[Leandro Regueiro]

Build of what? TM files? Translations index? Static HTML pages?

[Jordi Mas]

Translation memories files. I updated the description for clarity.