Many to one relationsship between components

### Describe your suggestion

Hi,

**Request**
Is it possible to support many-to-one relationsship between components ?

**Explanation**
As of now its possible to take the output of one component and set it as input for as many components you like in a pipeline.
What I would like to do, is to take the output of an unknown amount of components and point them all to the same input of another component.

**Why is this needed**
When creating ensemble-models I would like to test with different kinds of base-models. Azure ML could be great for this due to its pipeline + component + designer functionality. But to make this work, I need a component that takes either the forecast from multiple models or the trained model itself. See picture below for clearification:

With this functionality we could
1. Developers create a library of model-components (untrained models)
2. Data scientists can use the designer tool to drag and drop these models into a pipeline and create an ensemble of them using the ensemble component.

**Possible alternatives**
If one could define dependencies between components (e.g. combonent D shal not run before all of the components A to C are done), then this could also be solved by writing models/forecasts to a storage then using the same storage location as input in the ensemble component.

### Additional details

_No response_

Begin with the Azure ML pipeline, component, and Designer behavior described in the issue; no repository files or tests are identified. Done would require a documented implementation path for feeding an unknown number of component outputs into one input, or for using the proposed dependency and storage alternative.

Written by the indexing model from the issue text.