Discover subtitles in parallel to improve speed

[stil]

I used a stopwatch to measure how long it takes to discover subtitles for PL/EN and time was almost 23 seconds!

This request is pretty simple. Currently subtitles are discovered sequentially, one after another.

In result, it takes much longer than it should, we could to this in parallel and make multiple network requests at once.

Speedup would be pretty significant I suppose, at least couple of times faster is my bet.