[Task]: add context about using Group By in distributed environments (Beam)

### What needs to happen?

the Group By requires sorted input. In a distributed environment (like the Beam run configurations), sorting happens per node, which may lead to unexpected results.
This should be made more clear in the docs, we should point users to the Memory Group By transform, which doesn't require sorted input.

### Issue Priority

Priority: 2

### Issue Component

Component: Beam, Component: Documentation, Component: Transforms

Locate the documentation for the Group By and Memory Group By transforms, then review how Beam run configurations are described. Clarify that Group By requires sorted input and that distributed sorting is per node, and point readers to Memory Group By as the alternative; done when the limitation and alternative are clear in the relevant docs.

Written by the indexing model from the issue text.