More diagnostic messages are needed

[dmpetrov]

It is not easy to understand what happened if my job stocked. It would be great to have more logs.

[0x2b3bfa0]

Related to #70

[0x2b3bfa0]

What else can we do to display insightful diagnostic messages on failure?

[0x2b3bfa0]

Opposite of #154?

[0x2b3bfa0]

I guess that this issue can be safely closed. The new iterative_task resource already provides useful logs, and the legacy iterative_cml_runner resource is considerably more verbose than in November 2020. See iterative/cml#938 for further action points.