ML examples improvements, follow-up to IGNITE-9297

After IGNITE-9297 was resolved, ML examples were discussed with some folks (skozlov, avolkov, chief) in order to check if some other improvements may be desired. This ticket is to address obtained feedback.

List of points that look worth taking care of includes (but is not limited to):

• some examples javadocs still looks like missing details (eg random forest)
• some mentions of algorithms (eg kNN, gradient boosting etc) are better to be linked to respective introductory articles in order to help less experienced users easier grasp the examples
• presence of some examples doesn't look justified (eg LocalDatasetExample)
• logging in some examples looks sub-optimal, either too brief or too lengthy