Ansible Digital Ocean APIv2 too fast?

Question

I'm getting mixed results here provisioning new droplets with some new code. It seems like maybe Ansible is not waiting properly for the server to get an IP address. I have made several attempts to fix it with no luck and putting in some arbitrary wait time will not scale. Can anyone see if I made a mistake in the code that is causing this weird behavior. Right now, the playbook is designed to take the hostnames from the PROVISION group, spin up VMs, reply back with their IP addresses, and then append the new addresses to the DROPLETS group. Usually only part of the IP addresses make it into the DROPLETS group.

Here is the hosts file:

[DIGITALOCEAN]
localhost ansible_connection=local

[PROVISION]
TEST1
TEST2
TEST3

[DROPLETS]
TEST1 ansible_ssh_host=x.x.x.x
TEST2 ansible_ssh_host=y.y.y.y
See, this one didn't make it in... Not always this one... and they get unordered sometimes.

Here is the playbook:

---
- hosts: PROVISION
  gather_facts: no

  vars:
    do_token: TOKEN

  tasks:
  - name: Create new droplet
    digital_ocean: >
      state=present
      command=droplet
      name={{ inventory_hostname }}
      unique_name=yes
      size_id=512mb
      region_id=nyc3
      image_id=ubuntu-18-04-x64
      ssh_key_ids=KEY
      api_token={{ do_token }}
      wait=yes
      wait_timeout=500
    register: hostname
    delegate_to: localhost

  - name: Add host to Ansible config
    lineinfile:
      dest: "/etc/ansible/hosts"
      insertafter: '^\[DROPLETS\]'
      state: present
      line: "{{ hostname.droplet.name }} ansible_ssh_host={{ hostname.droplet.ip_address }}"
    delegate_to: localhost