mlr3batchmark 0.2.0

• feat: The design of batchmark() can now include parameter settings.
• feat: reduceResultsBatchmark gains argument fun which is passed on to batchtools::reduceResultsList.
  Useful for deleting model data to avoid running out of memory.
  Thanks to Toby Dylan Hocking @tdhock for the PR (#18).
• docs: A warning is now given when the loaded mlr3 version differs from the mlr3 version stored in the trained learners.
• feat: Support marshaling.
• feat: A renv project can be passed to batchmark() that is loaded in the job environment.