Use ontology id to lookup label

[theathorn]

The label values for e.g. library_preparation_protocol.library_construction_method (see HumanCellAtlas/dcp2#13) contain slight variations in the text used, which are then visible in the Data Browser. Indexing by looking up the ontology id would eliminate this issue.

[theathorn]

@hannes-ucsc to come up with a spec.

[hannes-ucsc]

The birds-eye view is to download and parse the HCAO ontology and convert a mapping at build time. The index loads the mapping and translates every term ID found in the metadata to the corresponding label. If no label is found in the mapping, the indexer logs a warning and falls back to the label / text in the mapping. Our metadata API needs to be extended to expose .ontology, .text and .label instead of eagerly choosing one of them. None of this affects the Azul API or the browser. The only externally visible effect will be that the service response contains the latest labels instead of the stale ones from the metadata.

[hannes-ucsc]

Converted this to an epic and added first child issue. Will add more child issues as the first child nears completion. There are some TBDs that I can only decide after we get more expertise with RDF, OWL and OBO.

[hannes-ucsc]

De-prioritizing given the changed priorities wrt AnVIL.