Ignore MasterTimeout for serverless deployment

[johnvanhienen]

According to the docs, the master_timeout attribute should be ignored when creating a serverless deployment. However, this isn't currently happening in practice. When trying to create an index, the apply fails with this error:

│ Error: Unable to create index: fietslanduo
│
│   with module.elasticstack["test.fietsplan"].elasticstack_elasticsearch_index.this["fietslanduo"],
│   on modules/ec-elasticstack/indices.tf line 2, in resource "elasticstack_elasticsearch_index" "this":
│    2: resource "elasticstack_elasticsearch_index" "this" {
│
│ Failed with: {"error":{"root_cause":[{"type":"status_exception","reason":"Parameter validation
│ failed for [/fietslanduo]: The http parameter [master_timeout] (with value [30000ms]) is not
│ permitted when running in serverless mode"}],"type":"status_exception","reason":"Parameter
│ validation failed for [/fietslanduo]: The http parameter [master_timeout] (with value
│ [30000ms]) is not permitted when running in serverless mode"},"status":400}

I'm proposing to handle this parameter the same way it's handled as WaitForActiveShards, which is already being properly omitted in serverless deployments. I've tested this fix by building the provider locally, and it works as expected.

[cla-checker-service]

💚 CLA has been signed

[tobio]

Thanks for this!