Add logger

[ablaom]

To close JuliaAI/MLJ.jl#1029.

This will be very similar to what we did for MLJTuning in these two PRs:

• Adding loggers into TunedModels MLJTuning.jl#193
• Make the global default_logger() the default logger in TunedModel(logger=...) MLJTuning.jl#221

That is: We add a logger field to the wrapper structs (there are two of them - ProbabilisticTunedModel and DeterministicTunedModel ) just like we did for TunedModel. And then we pass logger on to the Resampler instances , which in this case appears here. In this way, whenever any iteration control (e.g., WithLossDo) accesses a loss, it will also get logged to the specified logger.

Instead of nothing the default value of logger will be MLJBase.default_logger().

cc @pebeto