[BUG] apyhiveapi.heating/hotwater log a bare "'mode'" ERROR every poll while a device is offline (3,005 in 13 hours)

[peshmann]

Describe the bug

While a device is offline, apyhiveapi.heating and apyhiveapi.hotwater log an ERROR on every poll, and the message is just the bare key name — 'mode'. No device name, no context, no traceback.

I collected 3,005 of each (6,010 lines) in a single 13-hour outage, at one pair every 15 seconds:

2026-08-19 00:51:55.208 ERROR (MainThread) [apyhiveapi.heating] 'mode'
2026-08-19 00:51:55.209 ERROR (MainThread) [apyhiveapi.hotwater] 'mode'
2026-08-19 00:52:10.210 ERROR (MainThread) [apyhiveapi.heating] 'mode'
...

The cause is heating.py:

try:
    data = self.session.data.products[device["hiveID"]]
    state = data["state"]["mode"]
    if state == "BOOST":
        state = data["props"]["previous"]["mode"]
    final = HIVETOHA[self.heatingType].get(state, state)
except KeyError as e:
    _LOGGER.error(e)          # <- logs only the missing key name

_LOGGER.error(e) on a KeyError renders as the repr of the key, so the operator sees 'mode' and nothing else. hotwater.py has the same shape, and device_attributes.py has a third instance.

Two smaller points worth noting:

1. The try covers both data["state"]["mode"] and data["props"]["previous"]["mode"], so the log line cannot tell you which one was missing.
2. helper/hive_helper.py errorCheck() already has a dedupe mechanism (session.config.errorList) so that "Device offline could not update entity" is logged once per device rather than every poll. The getMode/getState handlers don't use it, so a known-offline device still generates ERROR-level noise indefinitely.

To Reproduce

1. Have a device go offline such that its product node no longer carries state.mode. (Mine came back from the API with error: MALFORMED_NODE_DESCRIPTOR, so the product was skipped and the cached entry had no state.)
2. Leave HA running.
3. Watch apyhiveapi.heating / apyhiveapi.hotwater emit an ERROR every 15 s indefinitely.

Expected behaviour

A known-offline or malformed device should log once, at WARNING, with enough context to act on — something like:

except KeyError as e:
    _LOGGER.warning(
        "Heating %s: product data missing key %s (device offline or malformed node)",
        device.get("haName", device.get("hiveID")), e,
    )

and ideally routed through the same errorList dedupe that errorCheck() already uses, so it doesn't repeat every poll.

Additional context

This matters beyond tidiness: at ERROR level and 5,760 lines/day it dominates the log, pushes genuinely useful entries out of the retained window, and trains people to ignore red text. It's also actively misleading — a bare 'mode' gives no indication that the underlying condition is simply "this device is offline", which the library already knows and already reports properly elsewhere.

Versions

• Home Assistant Core 2026.8.2 (HAOS)
• Hive-Custom-Component 2026.8.0
• pyhive-integration 1.0.9