add lifecycle to avoid forcing new vms

[hassenius]

Currently when doing a plan/apply on existing cluster the AWS provider might force rebuilding of nodes with something like

.aws_instance.icpnodes[2] (new resource required)
      id:                                                "i-0c3ef0428115d26cb" => <computed> (forces new resource)
      ami:                                               "ami-0cbf7a0c36bde57c9" => "ami-0cbf7a0c36bde57c9"
      arn:                                               "arn:aws:ec2:eu-west-1:675801125365:instance/i-0c3ef0428115d26cb" => <computed>
      associate_public_ip_address:                       "false" => <computed>
      availability_zone:                                 "eu-west-1c" => "eu-west-1c"
      cpu_core_count:                                    "4" => <computed>
      cpu_threads_per_core:                              "2" => <computed>
      ebs_block_device.#:                                "2" => "1"
      ebs_block_device.3042545595.delete_on_termination: "false" => "false"
      ebs_block_device.3042545595.device_name:           "/dev/xvdbc" => "" (forces new resource)

as per hashicorp/terraform-provider-aws#2036 ignore ebs_block_device to fix this