[Bug] `elasticstack_kibana_data_view` forces replacement due to auto-calculated `field_attrs`

[jalogisch]

The elasticstack_kibana_data_view resource is unexpectedly forced into a replacement state by Terraform. This occurs after the data view has been used in Kibana, which generates field popularity statistics. The provider detects these server-side generated statistics (specifically the count attribute within field_attrs) as a change from the state file, triggering a plan to replace the entire resource.

To Reproduce
Steps to reproduce the behavior:

1. TF configuration used:

   terraform {
     required_providers {
       elasticstack = {
         source  = "elastic/elasticstack"
         version = "~> 0.11.0"
       }
     }
   }
   
   provider "elasticstack" {
     // Elasticsearch & Kibana connection details
   }
   
   resource "elasticstack_kibana_data_view" "test_view" {
     data_view = {
       title = "my-test-logs-*"
       name  = "My Test Data View"
       id    = "my-test-data-view-id"
     }
   }

2. TF operations to execute to get the error:

   • Run terraform apply to create the resource.
   • In the Kibana UI, navigate to Discover and interact with the "My Test Data View" to generate field usage statistics.
   • Run terraform plan.

3. See the error in the output: The plan will show that the resource needs to be replaced.

     # elasticstack_kibana_data_view.test_view must be replaced
   -/+ resource "elasticstack_kibana_data_view" "test_view" {
         ~ data_view = {
             - field_attrs     = {
                 - "host.hostname" = { # forces replacement
                     - count = 5 -> null
                   },
                 - "event.action" = { # forces replacement
                     - count = 12 -> null
                   },
               } -> null
               id              = "my-test-data-view-id"
               name            = "My Test Data View"
               # (3 unchanged attributes hidden)
           }
         ~ id        = "default/my-test-data-view-id" -> (known after apply)
           # (2 unchanged attributes hidden)
       }

Expected behavior
After the initial creation, subsequent terraform plan operations should show no changes to the infrastructure. The provider should ignore server-side, auto-calculated attributes that are not managed within the Terraform configuration.

Debug output
(To be provided after running the reproduction steps with TF_LOG=trace enabled)

Versions (please complete the following information):
 - OS: Linux
 - Terraform Version: v1.x
 - Provider version: v0.11.0
 - Elasticsearch Version: 8.x

Additional context
The root cause appears to be that the provider does not differentiate between user-managed attributes and dynamic, server-side attributes like field popularity counters. This results in a perpetual diff, making idempotent management of the resource impossible once it is actively used.