fix: ranger did not receive num.threads = 1 during predict()

[jemus42]

Here

mlr3extralearners/R/learner_ranger_surv_ranger.R

Lines 118 to 120 in 0b3c766

	pv = self$param_set$get_values(tags = "predict")
	newdata = ordered_features(task, self)
	prediction = predict(object = self$model, data = newdata)

the parameters were retrieved, but never actually passed to predict().

This caused an issue in a benchmark setting where the prediction step caused a large spike in CPU load, and after asking Marvin he confirmed that prediction was multi-threaded in ranger, so this PR should fix that.