2 potential bug fixing or improvements

[Bustami]

Hey guys, thanks a lot for this great repo.

Last days I was working in a new github repo about passing networks in R, for my project Dato
Futbol. So, I adapted/translated a lot of your ideas from Python to R. Along the process i found 2 problems, related to Metrica Sport's tracking data, which you could consider as bug fixing or improvements.

1.- Due the data codification and/or its accuracy, I think it is necessary to consider some categories from the Subtype features as additional possesion changers, besides "BALL LOST" & "BALL OUT", in order to get correctly the on ball and off ball frames.

I'm using "HEAD-ON TARGET-GOAL", "ON TARGET-GOAL", "WOODWORK-GOAL", "OFF TARGET-OUT", "HEAD-OFF TARGET-OUT", "ON TARGET-SAVED" as well.

2.- Even though you define a function to inverse the Y axis named "to_metric_coordinates", I'm afraid you are not calling it along the code running.

Cheers