v0.1.21 – Entity cleanup and rename

What's New

• Entity cleanup – Removed redundant sensors (SWC Output %, SWC Low Output %, Error Code Text) and the Power switch (not controllable via API).
• Renamed entities for consistency: "Chlorinator" (binary sensor), "Chlorinator Output / Low Output / Low Mode" (numbers/switch), "Error", "Authentication", "MQTT Connected", "Device Online".
• Aux 2 switch hides automatically when Aux 2 is configured as a heat pump (climate entity takes over).
• Refresh Interval moved from a number entity to the integration's Configure dialog (Settings → Devices & Services → Exo Pool → Configure).
• MQTT Connected sensor now reflects the actual MQTT client state.
• Authentication sensor is always available (was incorrectly gated on coordinator data).

Note: After updating, existing entity names will change. You may need to update dashboards and automations that reference the old names.

v0.1.20

What's new

State bounce fix - After sending a command via MQTT (e.g. turning off the chlorinator), the UI no longer flips back to the old state.

AWS IoT echoes back an intermediate shadow update immediately after a write, but the reported section still reflects the old device state (the device has not processed the command yet). That stale update is now ignored during the 5-second post-write settling window, so the optimistic state stays correct until the device confirms the change.

v0.1.19

Extract auth.py; slim api.py to 836 lines; bump to 0.1.19

- auth.py: _full_login, _refresh_token, _store_aws_credentials,
  _refresh_authentication — all with lazy imports from api.py
- api.py: re-exports auth symbols; down from 1298 → 836 lines across
  three extraction rounds (write_manager, coordinator, auth)
- manifest: 0.1.18 → 0.1.19

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

v0.1.18

Switch improvements & batch schedule updates

• Switch toggles (Aux, Chlorinator, Power, …) send instantly — no more 55-second post-write block
• Failed switches revert their state and show a HA error notification; one automatic retry on network errors
• New service exo_pool.set_schedules to update all schedules in a single API call (~45s instead of several minutes)