[chrisism]

And all the commands gave me issues or nothing, except .... the os.system(..) as you mentioned @malte .
At least it started Retroarch with me. Still have to check if all the parameters still work.
@Wintermute0110  I'll go and make a branch for this, to separate it from the linux calls and test it out further. Looks like we have a winner. 

 Do you happen to have the debug log from this test script?  I'd be interested in what errors, if any, are thrown when it doesn't work. 

As requested omosohunoz (paste)
So, this is a clean run with that block of test code. I added an altered original subprocess.call before that. I did not open Retroarch beforehand and I do not have my Shield rooted.
As you see the last call actual executes and I can confirm that it indeed opened up Retroarch with the actual rom.
Also the launcher.log file which all the stdout is written to is completly empty. Ofcourse os.system does not have an option for stdout parameters anyway.