Bug 985: Correct graph query in apstra_datacenter_device_allocation resource when multiple candidate interface maps exist

[chrismarget-j]

Issue #985 describes a scenario where a given system node has multiple interface_map nodes linking it to the device_profile node which matches the serial number of the user-specified switch.

Consider the following terraform configuration:

resource "apstra_datacenter_device_allocation" "issue_985" {
  blueprint_id             = "irrelevant"
  node_name                = "A"
  initial_interface_map_id = "IF_MAP_A"
  device_key               = "123456"
}

along with the following graph elements:

                                /-->(interface_map IF_MAP_A)--\
(system A)-->(logical_device)==                                 ==>(device_profile B)
                                \-->(interface_map IF_MAP_B)--/

In this case, the user has identified:

• the system (A)
• the interface_map (IF_MAP_A)
• the logical_device (by specifying device key 123456 which is a model "B" device)

The correct graph traversal is obvious, but the code was not making use of the supplied interface map ID when querying the graph.

As a result, we'd find multiple matches and produce an error if multiple paths are found between the system and the device_profile nodes.

We should have been checking all of the user-specified details, both to avoid ambiguity, and to validate the user input.

This PR adds the initial_interface_map_id attribute to the graph query if the user has supplied it.