Errors in worker container near-impossible to debug.

[RPCMoritz]

When switching our environment from an admin user to a regular user, we ran into this, in our container logs:

Container: container_e11706_1594030029461_0001_01_000038 on host_8099
================================================================================                                                                   ==============
LogType:dask.worker.log
Log Upload Time:Mon Jul 06 15:37:11 +0200 2020
LogLength:69
Log Contents:
.skein.sh: line 2: environment/bin/python: No such file or directory
End of LogType:dask.worker.log

Repeated a hundred times.
I've been trying to trace down the actual contents of .skein.sh, and I assume it's merely the conda activate in line 1 of that file failing. But I have no idea, and it would be nice if the script could dump itself into the logs if it fails, so the actual values can be examined for reproducing the issue step-by-step.
This probably requires adding a finalize method to the script, which dumps its own contents, when the script goes into an exit 1 condition.