added an initial approach for the text conversion

[splitbrain]

This is completely untested and is mainly here as a reminder on what to do when we implement media indexing. We want a way to:

• use local cli tools
• use a remote HTTP service
• have support for additional metadata reading
• configure different parsers for different extensions
• guess the language

Still needed:

• tests
• config loading
• check if we want any additional meta data
• maybe use exceptions instead of return codes
• everything else to actually use this