random forest: providing training job with aggregator job id fails

[jdries]

When referencing a random forest by job id, this doesn't work:
training_job = c.job("vito-j-cc66a593416043468e4c5ca83ef90b28")
it gives an error when trying to apply random forest.

This does work:
training_job = c.job("j-cc66a593416043468e4c5ca83ef90b28")

probably the 'vito-' prefix needs to be stripped automatically.

[soxofaan]

FYI: the vito- prefix is added to the job id by the aggregator (to be able to determine which back-end a given job id belongs to).

If that job-id is used as reference for loading a ML model, the aggregator should strip it away again when forwarding to the upstream back-end. So I think this is an aggregator bug.
For example, we had to do the same in load_result before: #19

[soxofaan]

done