For Each Issue

You can do it without using a loop using something like the below; it's a bit difficult to know if this works as you've not given a structure for your CDTs or any sample data, but this should point you in the right direction.

index(
  ri!testCdt,
  max(
    intersection(
      wherecontains(
        ri!testCdt.aId,
        ri!tData_cdt.id
      ),
      wherecontains(
        ri!testCdt.tabId,
        ri!tData_cdt.tName
      )
    )
  ),
  null
)

The one that doesn't use a loop should be faster - a!forEach can be a bit slow depending on the amount of data involved. However, you should be able to answer this question yourself using the performance view as documented here. :)