refactor: use the new flow/flownode links from the bpmn semantic in the predictions demo

[csouchet]

Live environment of examples

https://cdn.statically.io/gh/process-analytics/bpmn-visualization-examples/476-Use_the_new_flow/flownode_links_from_the_BPMN_Semantic_in_the__Predictions__demo/demo/predictions/index.html

last commit: https://cdn.statically.io/gh/process-analytics/bpmn-visualization-examples/2eebd856f88103797eabea27035cc309e23f2d6a/demo/predictions/index.html

Closes #476

[tbouffard]

I cannot review this PR until #514 is merged. There are commits coming from the dependent PRs and new commits in this one.
This make the review hard to do. So I will wait for the merge of the other PRs.

[tbouffard]

issue: we cannot duplicate the code from the demo.
We already know it requires refactoring at least for process-analytics/bonita-day-demo-2023#52. There are also a lot of extra validation of the incoming/outgoing (verification that a related edge id exists in the model) whereas this should be guarantee by bpmn-visualization.
In addition, we plan to extract PathResolver from the bonita day demo process-analytics/bonita-day-demo-2023#36. If we duplicate the implementation here, we must track it and plan to use the shared code here. And this may not possible out of the box depending on the way we package the shared code. In particular, we are not sure to to provide an IIFE bundle for the shared lib.

[csouchet]

To replace in another PR. See #260

[tbouffard]

Let's talk about alternatives to the code duplicated from the bonita day demo (PathResolver).
As I mentioned in #514 (review), it's becoming increasingly clear to me that we should consider moving this demo to its own repository to benefit from the package manager, the shared library and the framework.

[tbouffard]

After discussions with @csouchet, we agreed to merge this PR as it is today.
#260 includes a task to use the shared code of PathResolver when available.
We also agreed to review the extraction of this demo in a dedicated repository later.