Plain indexing takes too long

[RodrigoHolztrattner]

Visiting and opening each file takes a huge slice of time, there are better ways to approach this:

1 - Only index what was used/requested
2 - Index all superficial data (paths + filenames) but only index internal data (icon + references) when the file is requested or modified
3 - Asynchronously do this processing (not a good idea, would take a long time anyway)
4 - Use a cache and one of the approaches below, I believe a mix of this and the option 2 would be the best idea

This fix shouldn't imply in any external changes for existing projects so any ongoing development with the updated branch can continue without problems.