Pipeline defined only by name may put burden on pipeline owner

[nerdstrike]

Our definition for "Pipeline" is name, uri and version, but only the name is constrained as unique - this way the same code can be used for different purposes and parameterisations. There are some possible consequences:

Scenario: A minor change in a given pipeline results in a change in the version.
Assumption: We should not mix/modify existing pipelines for task-provenance reasons
Result: The pipeline owner must register the same pipeline again with the new version and a new name. This forces the pipeline owner to do ad-hoc versioning via naming, and do so wisely.
Example: name="spaceranger", name="spaceranger+fix", name="spaceranger v2" and so on

It is expected on a per-pipeline-execution basis that the outputs will include sufficient provenance on what was run. However, for the purposes of reanalysing prior runs (as in the case of essential pipeline/protocol fixes), do we need to distinguish between versioned executions?