Refactor the discovery.xml fetching

[donquixote]

We have the CoolRequest class which fetches the discovery.xml and extracts the WOPI client url from the xml. I would like to refactor this.

Refactoring:

• Replace global functions with services.
• Avoid object state where it is not needed.

Functional changes:

• Rethink error handling.
• Don't concatenate t() strings, and rethink error message passing.
• Use Guzzle instead of file_get_contents to fetch the remote content.
• Cache the result, to avoid repeated calls. This also means we need to flush that cache at some point.
• Allow getting different values from the xml, e.g. for different mime types.

@hfiguiere some questions. In CoolRequest we have these error codes 101, 201 etc. Do these numbers have special meanings? Is there any external system that understands these number codes?

[hfiguiere]

In CoolRequest we have these error codes 101, 201 etc. Do these numbers have special meanings? Is there any external system that understands these number codes?

They come from our PHP example (I didn't write that):

https://github.com/CollaboraOnline/collabora-online-sdk-examples/blob/master/webapp/php/load.php

I also reuse them in the WordPress plugin.

We could use different values if that's better.