cwlprov:relationship sketch

Together with #1 this attempts to find a way to pre-define domain-specific provenance that would be generated at workflow run time. The idea is define a set of relationships that will be added onto the produced outputs of a step to relate it to other data values or concepts at creation time.

These can use domain-specific ontologies like [EDAM ontology](http://edamontology.org/page) or [BioSchemas](https://bioschemas.org/specifications/), or more generic ones likes [PROV](https://www.w3.org/TR/prov-o/) or [schema.org](https://schema.org/)

``` cwl
#!/usr/bin/env cwl-runner
cwlVersion: v1.0
class: Workflow

inputs:
first_input: File
second_input: long

steps: []

outputs:
first_output:
type: File
outputSource: first_input
cwlprov:relationships:
prov:wasDerivedFrom: [ '#inputs.second_input' ]
prov:wasInfluencedBy: [ '#inputs.second_output' ]

$namespaces:
prov: http://www.w3.org/ns/prov#
cwlprov: https://w3id.org/cwl/prov#

$schemas:
- http://www.w3.org/ns/prov.owl
```

Start by reading issue #1 and the provenance relationships shown in this issue's CWL example, including the referenced PROV, EDAM, BioSchemas, and schema.org material. Define the intended relationship set and how those relationships should be added to workflow outputs at run time; no implementation files or tests are named, so completion criteria need to be established first.

Written by the indexing model from the issue text.