Currently when we are analysing a document within an array, information about the document does not get aggregated.

Instead,
every key value pair in a document should get it's own instance of schema_parser (SchemaParser::new() and not finalise_schema())