feat: add leaf_min_events_ratio for surv.aorsf

[jemus42]

Based on #383

This adds leaf_min_events_ratio to surv.aorsf to allow tuning leaf_min_events as a ratio, analogously to mtry[._]ratio parameter in this and other forest-based learners.

I have settled for an intermediate upper bound of 0.5, corresponding to half the number of observed events in the task.
This is most likely a bit confusing, but it corresponds to what the upstream learner enforces.

I'm not sure what a sensible conversion is here, considering the following example:

library(mlr3)
library(mlr3proba)
library(mlr3extralearners)
library(mlr3pipelines)
library(mlr3tuning)
#> Loading required package: paradox

set.seed(123)
task = tsk("lung")


# ratio of 0 yields a minimum of 1, sensible?
lrn1 = lrn("surv.aorsf", leaf_min_events_ratio = 0)$train(task)
lrn1$model$leaf_min_events
#> [1] 1

# A reasonable case, I think?
lrn2 = lrn("surv.aorsf", leaf_min_events_ratio = 0.3)$train(task)
lrn2$model$leaf_min_events
#> [1] 37

# This is a bit on the edge?
lrn3 = lrn("surv.aorsf", leaf_min_events_ratio = 0.49)$train(task)
lrn3$model$leaf_min_events
#> [1] 60

lrn4 = lrn("surv.aorsf", leaf_min_events_ratio = 0.50)$train(task)
#> Error: leaf_min_events = 61 should be <= 60 (number of events divided by 2)

# Here the param set restriction hits, maybe confusingly so?
lrn5 = lrn("surv.aorsf", leaf_min_events_ratio = 0.51)$train(task)
#> Error in self$assert(xs, sanitize = TRUE): Assertion on 'xs' failed: leaf_min_events_ratio: Element 1 is not <= 0.5.

^{Created on 2024-09-10 with reprex v2.1.1}

[sebffischer]

Is this ready to be reviewed or still a draft?

[jemus42]

Pending some discussion about the API in #383 --- feel free to chime in, because I'm still uncertain what we want to do here.