Remove forced injection of `eval_time` in `predict_survival()`

[hfrick]

We add eval_time to all prediction calls in predict_survival.model_fit() and predict_hazard.model_fit() which causes problems if the the engine's predict method does not have an eval_time argument and does not swallow it via the dots, e.g. by enforcing empty dots.

The "aorsf" engine for rand_forest() in censored enforces empty dots which is what lead to tidymodels/censored#236. The dev version of censored does not rely on this pattern anymore so once that is on CRAN we should remove those lines:

parsnip/R/predict_survival.R

Lines 37 to 38 in 8f13c1c

	# Pass some extra arguments to be used in post-processor
	object$spec$method$pred$survival$args$eval_time <- eval_time

parsnip/R/predict_hazard.R

Lines 35 to 36 in 8f13c1c

	# Pass some extra arguments to be used in post-processor
	object$spec$method$pred$hazard$args$eval_time <- eval_time

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.