fix fallback for fit when there is a target supplied

oops

Author: ablaom

src/model_api.jl

@@ -16,7 +16,11 @@ fit(::Static, ::Integer, data...) = (nothing, nothing, nothing)
 fit(m::Supervised, verbosity, X, y, w) = fit(m, verbosity, X, y)
 
 # fallback for unsupervised detectors when no "evaluation" labels appear:
-fit(m::Probabilistic, verbosity, X, y) =  fit(m, verbosity, X)
+fit(m::Union{AbstractProbabilisticUnsupervisedDetector,
+             AbstractDeterministicUnsupervisedDetector},
+             verbosity,
+             X,
+             y) =  fit(m, verbosity, X)
 
 """
     MLJModelInterface.update(model, verbosity, fitresult, cache, data...)