Problem while selecting the correct error event subprocess

**Describe the bug**
I have a BPMN with a subprocess containing another subprocess.
In both subprocesses, i can throw BPMN Errors.
I added, in each subprocess, an error event subprocesses to handle the BPMNError occuring inside.

When the engine search for an appropriate error subprocess to handle a BPMNError, the selected one is not always the same.

I put a full explanation of the problem on the flowable forum :
https://forum.flowable.org/t/bpmnerror-error-handling-subprocess-possible-bug-at-least-strange-behaviour/10873

**Expected behavior**
If i have a BPMN containing ITEMSubprocess containing a **sub**ITEMSubproces expect that :

- A BPMNError happening in a task of ITEMSubprocess will be handled by the error event subprocess of ITEMSubprocess .
- A BPMNError happening in a task of *sub**ITEMSubproces will be handled by the error event subprocess of *sub**ITEMSubproces.

**Code**
https://github.com/flowable/flowable-engine/blob/6c5b934c0b0fa8b33a3612a11f503c06b81a369a/modules/flowable-engine/src/main/java/org/flowable/engine/impl/bpmn/helper/ErrorPropagation.java#L155C1-L175C18

I think the issue is caused by the fact that :

`List events = eventMap.get(refId);
for (String refId : eventMap.keySet()) { //Not always in the same order ; probably due to hashing`

`currentContainer.getFlowElement(refActivityId) // considers that a refActivityId in subITEMsubprocess is "contained" within ITEMSubprocess.`

==> I get inconsisted results depending on the order of the eventMap keySet;
The selected error handling subprocess is the one of the last checked subprocess between ITEMsubprocess and subITEMsubprocess.
I think it should search for an error handling subprocess only withing current subprocess and it's parents; not in children subprocesses.

**Additional context**
flowable 7.0.0

Empieza en modules/flowable-engine/src/main/java/org/flowable/engine/impl/bpmn/helper/ErrorPropagation.java alrededor de las líneas 155-175 y después lee la explicación enlazada del foro para el caso de subprocesos anidados. Reproduce el manejo de errores con subprocesos anidados y distintos órdenes de eventMap; se considera terminado cuando los errores de cada subproceso son gestionados por ese subproceso o por un parent contenedor, nunca por un child.

Escrito por el modelo de indexación a partir del texto del issue.