[core] improved error reporting during env deployment/configuration

[knopers8]

• "task '/opt/o2/bin/o2-readout-exe' on alio2-cr1-mvs11 (id 2sBwA3Z8yWU, name alio2-cr1-hv-gw01.cern.ch:/opt/git/ControlWorkflows/tasks/readout@12b11ac4bb652e1835e3e94806a688c951691d5f#2sBwA3Z8yWU) failed with error..." displayed by COG becomes "task 'readout' on alio2-cr1-mvs11 (id 2sBwA3Z8yWU) failed with error...". I believe the removed information is not useful for typical operators. Details can be always found in logs.
• "MesosCommand MesosCommand_Transition timed out for task 2sBwYFZ82wn" becomes "MesosCommand_Transition timed out for task 2sBwYFZ82wn" because by convention, all MesosCommands have "MesosCommand" in their names.
• If a transition request times out, more accurate error is reported in COG. "nil response" becomes "CONFIGURE could not complete for critical tasks, errors: task 'readout' on alio2-cr1-mvs11 (id 2sBwA3Z8yWU) failed with error: MesosCommand_Transition timed out for task 2sBwA3Z8yWU".
• In case a task transition times out, any other errors which happened at the same time (e.g. a task crashed during transition) are not omitted anymore in COG.
• "nil response" in task/manager.go becomes "no response from Mesos to CONFIGURE transition request within 120s timeout", but it will not be typically printed for tasks timing out during a transition.
• "nil response" in commandqueue.go becomes "did not receive neither response nor error for MesosCommand_Transition"

One could consider further simplification of some of these messages, but perhaps let's see how this goes.

OCTRL-975