TDoA3 outlier filter need update for Long Range

[krichardsson]

The TDoA3 outlier filter was originally written for small spaces and the "normal" mode, when running the Loco system in TDoA3 Long Range mode in a large system, it is not doing a great job resulting in outliers getting through. It is probably a good idea to look at the implementation and see if it can be improved, for instance also using the direction to the sample and not only the distance. The outlier filter should be re-written but it will be easier to do after the kalman estimator has been added to the python bindings (#602).

---

A temporary workaround that seems to solve some of the problem in a large system is to change the settings to

#define MAX_BUCKET_FILL 7
#define FILTER_CLOSE_DELAY_COUNT 20

Adding a maximum error distance that can be set from the ground when the filter has converged is also an option.
Conceptually something like the following code could be inserted here

  if (errorDistance > maxFilterDistance) {
    sampleIsGood = false;
  }