Well-known vocabularies are not completely deleted from visual model

[RadStr]

• When user deletes some well-known vocabulary and then readds it back, the old data about visibility is still present.
• Doesn't seem like that big of a deal, but it is probably something to be aware of.

This issue occurs for both packages stored in manager and those not stored

When user isn's working in package

• That is, the user opened the https://tool.dataspecer.com/conceptual-model-editor/diagram link
  

When user is working in package

That is user opened visual model from the package manager.

Since the GIF for this case is quite long (around 50 seconds), I will only include the steps in text form

1. Create package in package manager
2. Add visual model
3. Open Visual model
4. Add some well-known vocabulary and add some classes from it to canvas
5. Save package
6. (Optionally go back to manager and reopen the visual model)
7. Remove the well-known vocabulary from CME
8. Save the package
9. (Optionally you can go back to manager and open the visual model again to verify that the semantic model was actually removed and the problem isn't because there is something stored in browser memory)
10. (Reopen the visual model)
11. Readd the well-known vocabulary and see that the previously added classes are all still there

[RadStr]

Maybe related to #637 or rather #636 (that one talks about workspace export though), but since the visual model changed quite a lot, I thought that these are no longer a problem