Evaluate and maybe add optimizations to GCC implementation

[mengelbart]

There are a couple of optimizations that could improve the current GCC implementation which I'd like to test and add if they prove useful:

• [] Implement linear regression as an alternative to the kalman filter
• [] Improve efficiency of pkg/gcc/rtt_estimator.go:
  • [] Use a better ring buffer implementation
  • [] Calculate rate once and afterwards only update with new measurements
• [] pkg/gcc/rate_controller.go uses bitsPerFrame calculated by assuming 30 frames per second, can/should we make this configurable?