Report load averages of 0 instead of going unavailable

[shauneccles]

Problem

The load-average sensors' value_fn ended in … and la[N]) or None. When a load value is genuinely 0 — e.g. an idle router's load average — and la[N] evaluates to 0 (falsy), so 0 or None returns None and the sensor goes unavailable instead of reporting 0. All three load sensors (1m/5m/15m) share this.

Fix

Return the indexed value directly when the list is well-formed, falling back to None only when the data is actually missing/malformed:

value_fn=lambda system_status: (
    la[2]
    if isinstance(la := system_status.get("load_average"), list) and len(la) > 2
    else None
),

Validation against real hardware (GL.iNet MT6000, firmware 4.9.0)

Tested in a Home Assistant dev instance connected to a live MT6000, captured at the same condition (an idle router reporting a 0 load average).

Before (this branch's base, master) — router's 15-minute load = 0:

sensor.gl_inet_mt6000_load_avg_15m = unavailable      <-- bug

After (with this fix) — router fully idle, router_get_status() returned load_average: [0, 0.02, 0]:

sensor.gl_inet_mt6000_load_avg_1m  = 0                <-- reports 0 (was unavailable under the bug)
sensor.gl_inet_mt6000_load_avg_15m = 0                <-- reports 0 (was unavailable under the bug)

Deterministic check on the exact failing input:

load_average=[0.06, 0.02, 0]    -> OLD=None    NEW=0      <-- the fix
load_average=[0.05, 0.03, 0.01] -> OLD=0.01    NEW=0.01   <-- no regression
load_average=None               -> OLD=None    NEW=None   <-- still safe
load_average=[]                 -> OLD=None    NEW=None   <-- still safe

Pre-existing bug (present unchanged on master), independent of #150 — raising it on its own as discussed.