chore(release): version packages

[github-actions]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and publish to npm yourself or setup this action to publish automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to master, this PR will be updated.

Releases

forty-two-watts@0.106.0

Minor Changes

• 9638c78: Ferroamp self-healing watchdog for the sticky-pplim trap. When the
  SSO reports the post-incident signature — DC bus voltage > 200 V, zero
  PV current, no fault, relay closed — continuously for ten minutes, the
  driver now auto-publishes pplim arg=<pplim_release_w> to release
  the lock. Operator opts in by setting config.pplim_release_w > 0;
  without it, the watchdog logs a per-incident warning but does not
  publish (we have no safe release value to send).

  A five-minute cooldown between successive recoveries prevents command-
  spam if the release doesn't take. A new stuck_pv_recovery_count
  metric tracks lifetime recovery count so operators can alert on a
  chronic condition.

  Reuses the existing pplim_release_w field — same value, dual
  purpose (dispatcher curtail_disable release AND watchdog
  self-recovery).

  Layered with #367
  (driver hard-fail on pplim arg=0) and the dispatcher fix in the
  parallel PR (fix(curtail): no spurious release ...) this is the
  third and final layer of defense against the 2026-05-27 brick.

Patch Changes

• 312e9ba: Defense-in-depth against the 2026-05-27 Ferroamp brick. Two
  independent changes that, combined with PR fix(ferroamp): refuse pplim arg=0 — sticky-lock trap on extapi #367's driver-side hard
  fail on pplim arg=0, eliminate every known trigger path:

  • Dispatcher: ComputePVCurtail no longer emits a curtail_disable
    release simply because a previously-curtailed driver dropped out of
    the proportional allocation due to its own |PV| crashing to ~0
    (often a direct consequence of OUR curtail throttling that driver
    down). The release is now only sent when the curtail directive
    truly clears, or the driver is removed from SupportsPVCurtail, or
    the driver goes offline. Also: per-driver allocations rounding to
    ≤ 1 W are suppressed entirely — never publish a near-zero
    pplim that some inverters treat as a hard "limit to 0 W" lock.

  • Ferroamp driver: subscribes to extapi/control/response
    (was: extapi/result — wrong topic, never received anything),
    parses {"status":"ack|nak", ...} responses, and exposes
    cumulative extapi_nak_count + extapi_ack_count metrics. NAK
    responses are also logged as warnings with transId + msg
    fields. The 2026-05-27 brick was preceded by minutes of
    nak: no available ESOs detected in system that we couldn't see
    through ftw telemetry — now the operator can alert on any non-zero
    NAK rate.

  Tests added:

  • Four new dispatcher regressions in control/pv_curtail_test.go
    guarding the suppression / release semantics.
  • One driver test in drivers/lua_ferroamp_curtail_test.go
    asserting NAK + ACK counter advancement.

• 322ffe2: Ferroamp safety fix: the Lua driver now refuses to publish
  pplim arg=0 from any curtail / curtail_disable path.

  Ferroamp's extapi treats {"cmd":{"name":"pplim","arg":0}} as
  "limit PV output to 0 W" — same wire bytes as a naive release would
  have, opposite semantics. The inverter sticks at 0 W PV until the
  operator clears pplim from the Ferroamp portal or power-cycles the
  EnergyHub. On 2026-05-27 this fired against a live SE4 site after the
  dispatcher's proportional curtail allocation gave a 0-share to
  Ferroamp; recovery required a 30+ minute outage and a portal-side
  reset.

  Changes:

  • curtail with power_w <= 0 is now a logged no-op (was: published
    pplim arg=0).
  • curtail_disable is a logged no-op by default (was: published
    pplim arg=0). To restore automatic release, set
    config.pplim_release_w on the driver to the inverter's nominal
    max (e.g. 15000 for a 15 kW SSO). The driver then publishes
    pplim arg=<release_w> which Ferroamp accepts as "raise the limit".
  • New unit tests guard the wire payload against any regression that
    reintroduces pplim arg=0.
  • Docs in docs/configuration.md describe the trap and the new
    config field.

  Operators with supports_pv_curtail: true on Ferroamp should add
  config.pplim_release_w: <SSO-rated-watts> to keep curtailment
  auto-releasing. Without it, curtail still engages correctly, but
  release becomes a portal action.