Fix automatic data refresh of remote source

[noliveleger]

Description

When using pulldata feature in a project with a remote URL, data is now refreshed automatically.

Notes

To calculate the hash of a remote file, KPI and KoboCAT were using two different ways to achieve the goal.
It is now using the same utility which relies on remote server to expose specific headers (Etag, Last-Modified or Content-Length*) . If none of them are provided, a custom hash is built and cached (for a specific amount of time) to avoid too many hits on the remote server.

* Content-Length could be not accurate because two different files could have the same length.