Following changes to trusty-ai to embed kogito-tooling editors (currently DMN but eventually BPMN, PMML's Scorecards and Linear Regression models too) additional improvements have been identified.

This JIRA represents the mechanism to retrieve and pass the model XML. Currently a single REST endpoint (in trusty) returns the model XML and meta-data. This is parsed browser-side into a TypeScript object by axios before being passed to the ModelDiagram component. This extracts the XML `string` and passes it to the embedded editor. The embedded editor then passes the XML `string` across an IFRAME boundary before the editor converts the XML into an object model for rendering. Phew!

There are concerns that the parsing by axios could be a performance issue for large models.

Therefore the proposal is:

1. Expose the model meta-data as one endpoint, MIME=json/application
2. Expose the model XML as another endpoint, MIME=text/xml
3. Add the ability for the embedded editor to accept the endpoint URL from which to retrieve the XML thus eliminating the cross IFRAME passage of potentially large string objects.