Allow options to mount data volumes while running a job in sciserver-compute??

[umesh-timalsina]

[brollb]

This might be useful to have. My biggest concern is the implicit coupling between the pipeline and the compute backend (and possible specific settings).

How would these errors be handled in a way that is informative? For example, suppose I want to run someone else's pipeline and I don't know that specific operations require specific compute backends (and configuration settings)? Furthermore, what if operations require different compute backends or configurations?