[ML] Outlier detection method names `distance_kth_nn`, `distance_knn` are invalid

[dimitris-athanasiou]

When running outlier detection with method being set to either distance_kth_nn or distance_knn the process rejects those as invalid values.

"state" : "failed",
      "failure_reason" : """[test-1] Error closing data frame analyzer process [Input error: bad value '"distance_kth_nn"' for 'method'.]""",

[dimitris-athanasiou]

The reason this happens is because those methods were renamed without also following up the rename in the java side. However, the old names are more descriptive. I'll raise a fix to rename them back to the expected names.