Find a way to deal with search-metadata.json

the context https://bluebrain.github.io/nexus/contexts/search-metadata.json it is a generated context from all the `MetadataContextValue` that every plugin publishes.

However if someone creates a third party plugin, the content of the file https://bluebrain.github.io/nexus/contexts/search-metadata.json (published to github pages) won't change and someone relying on some json-ld library to resolve that context will probably have issues.

In order to fully support that, we will probably have to do the following. change:
```scala
final case class MetadataContextValue(id: Iri, value: ContextValue) extends AnyVal
```
So, instead of having just one aggregated context(search-metadata.json) we will have a set of contexts, and elasticsearch listings response should return them in the `@context` array. In this way we will support plugin extensibility.

Start with MetadataContextValue and trace how the aggregated search-metadata.json context is generated and how Elasticsearch listings responses assemble their @context. Compare the current published context with the proposed set of plugin contexts; the work is done when third-party plugin metadata contexts can be returned and resolved without changing the published aggregate.

Written by the indexing model from the issue text.