TEST-21: delivery_hold_s's floor re-derived, and 30.0 s did not reproduce

No code and no option changes. What changes is a number this repository claimed
about a shipped safety constant.

`delivery_hold_s: 185` shipped resting on "30.0 s, the worst continuous
legitimate episode at the shipped threshold across all 2640 runs", plus a lag
sweep giving 45.0 s and 90.0 s - from a harness `DECISIONS/SAFETY-42` §1
deliberately did NOT commit, because it imports `plantsim` and `frame_check` out
of admin/goodwe. So the floor under a safety constant could not be re-run. That
is PROC-01's shape, with eleven instances in three days.

The harness is committed now, in admin/goodwe as `sim/delivery_window.py`, and
NOT here. `plantsim` and `frame_check` are that repo's; restating them in this
tree is CI-08's coupling, which has broken three times, and the alternative
fate - skipped for want of its imports - is a green that establishes nothing.
There the dependency runs one way: THIS repo's `control`, `arbiter` and the
shipped `delivery.delinquent()` are imported live out of whichever checkout is
under test. Its `--selftest` is now a step in that repo's `checks.yml`, so a
change here that moves `saturation_w`, `max_w`, `delivery_hold_s` or the shape
of `delinquent()` goes red there instead of leaving a stale number here.

    cd goodwe/sim && GOODWE_ADDON_REPO=/path/to/goodwe-addon \
        python3 -B delivery_window.py --selftest --campaign --blag

⚠️ THE HEADLINE DID NOT REPRODUCE: 35.0 s, not 30.0 s. Recorded as measured. The
harness was not adjusted until it agreed - the one correction made during the
rebuild was made AGAINST the doc's number (§1 records it) and made every figure
LARGER.

WHAT REPRODUCED - and it is the part the compatibility promise rests on:
  * the N=1 control arm is 0.0 s at EVERY threshold in EVERY run. §3.2's algebra
    measured, not argued.
  * the flapping family is 0.0 s everywhere in all 840 runs - the family that
    defeated SAFETY-30's site window still cannot touch this check.
  * 45.0 s at twice A's lag, to the cycle.

WHAT DID NOT:
  * the headline, by one control cycle (30.0 -> 35.0 s); the reversal family by
    two (20.0 -> 30.0 s).
  * the NON-MONOTONE curve. §3.3 finding 3 read 55.0 s at 250 W against 40.0 s
    at 100 W as a property of the reduction. This rebuild's curve is monotone.
    The mechanism is real and is still the reason not to read either table as a
    sensitivity curve, but it is an ARGUMENT now, not a measurement, and it is
    written down as weaker.
  * the low-threshold rows, badly (40.0 -> 120.0 s at 100 W). Nothing ships
    there, but that gap is the honest measure of how much of the original table
    was a property of axes nobody wrote down. §3.3 fixes every axis COUNT and
    RANGE and no individual value; the rebuild prints its choices above its own
    table so the caveat travels with the number.
  * the four-times-slower row moved the OTHER way: 75.0 s, not 90.0 s.

185 s IS NOT CHANGED. It is 5.29x the re-measured worst (the record claimed
6.2x), 2.47x the four-times-slower case, 6.2x the firmware's LINK_STALE_S veto
and 4.9x inside the billing quarter. All four bounds clear and none narrowly,
and the ordering of the argument is unchanged. Moving a shipped safety constant
because a margin went from 6.2x to 5.3x while every bound holds would re-open
SAFETY-41's four-attempt false-trip regime for nothing.

⚠️ ONE CONSEQUENCE AN OPERATOR CAN HIT. At 30.0 s the rig's floor coincided with
the firmware's veto and a reader could not tell which was binding. At 35.0 s the
rig's is strictly the binding one - and it is ABOVE the schema's own
`int(30,900)` minimum. `delivery_hold_s: 30` is therefore below a MEASURED
legitimate episode and would report on ordinary saturation. Both translations
now say so. Raising the schema floor is a change to a shipped option's contract
and is left to its own card rather than smuggled in here.

⚠️ THE B-LAG ROWS ARE RELABELLED THROUGHOUT. Inverter B does not physically
exist (~2026-09-12); `plantsim.INVERTER_B` is A COPIED, `measured=False`. Every
two-inverter figure in this record is two copies of A, and the lag sweep is a
SENSITIVITY STUDY ON A HYPOTHESIS ABOUT B, not a measurement of it. The old
wording - "MEASURED, not felt" - overstated exactly that, and "the RAMP does not
move it at all" was too strong: halving it is worth 0 s at A's lag and +5 s at
twice it.

Touched: DECISIONS/SAFETY-42 §1 and §3.3, app/delivery.py's provenance comment,
config.yaml's `delivery_hold_s` block, both translations, CHANGELOG Unreleased.
26/26 suites pass.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BwRLw7eKYj2byVdwB6kbjU

goodwe_controller/config.yaml

@@ -560,17 +560,31 @@ options:
   # ⚠️ MEASURED ON THE TEST-19 TWO-MASTER RIG, NOT CHOSEN. 2640 closed-loop runs
   # at a roster of two in which nothing was faulty, over 6 gains x 5 slews - the
   # whole range this schema accepts - across load reversals, a flapping load and
-  # ASYMMETRIC SPLITS, plus 1890 runs at a roster of ONE as the control. At the
-  # shipped threshold the worst CONTINUOUS legitimate episode is 30.0 s (the
-  # weighted family; 20.0 s on reversals, 0.0 s on flapping, and 0.0 s at every
-  # threshold in all 1890 one-inverter runs). 185 s is 6.2x that.
+  # ASYMMETRIC SPLITS, plus a roster-of-ONE arm as the control. At the shipped
+  # threshold the worst CONTINUOUS legitimate episode is 35.0 s (the weighted
+  # family; 30.0 s on reversals, 0.0 s on flapping, and 0.0 s at every threshold
+  # in every one-inverter run). 185 s is 5.29x that.
   #
-  # ⚠️ AND THE MARGIN IS SIZED ON A MEASUREMENT, NOT ON A FEELING, BECAUSE
-  # INVERTER B DOES NOT PHYSICALLY EXIST. The rig's B is A copied. Sweeping a
-  # DIFFERENT second unit through that campaign: the worst legitimate episode is
-  # insensitive to B's ramp (30.0 s at half A's ramp) and scales with its LAG -
-  # 45.0 s at twice A's measured 10 s lag, 90.0 s at four times it. 185 s is
-  # 2.06x even the four-times-slower case.
+  # ⚠️ AND THE CAMPAIGN IS RE-RUNNABLE NOW, WHICH IT WAS NOT WHEN THIS NUMBER
+  # SHIPPED. TEST-21 (2026-09-07) committed the harness to the OTHER repository,
+  # `admin/goodwe:sim/delivery_window.py` - it imports plantsim and frame_check,
+  # which that repo owns. From clean checkouts of both, side by side:
+  #
+  #     cd goodwe/sim && GOODWE_ADDON_REPO=/path/to/goodwe-addon \
+  #         python3 -B delivery_window.py --selftest --campaign --blag
+  #
+  # ⚠️ THE ORIGINAL RECORD SAID 30.0 s AND IT DID NOT REPRODUCE - the re-run
+  # says 35.0 s. 185 s still clears all four bounds and is unchanged; what
+  # changed is the claim. DECISIONS/SAFETY-42-unit-delivery.md §3.3 carries both
+  # tables and what did and did not reproduce.
+  #
+  # ⚠️ AND THE MARGIN IS A SENSITIVITY STUDY ON A HYPOTHESIS, NOT A MEASUREMENT,
+  # BECAUSE INVERTER B DOES NOT PHYSICALLY EXIST (~2026-09-12). The rig's B is A
+  # COPIED. Sweeping a DIFFERENT second unit through that campaign: the worst
+  # legitimate episode barely moves with B's ramp (35.0 s at half A's ramp) and
+  # scales with its LAG - 45.0 s at twice A's measured 10 s lag, 75.0 s at four
+  # times it. 185 s is 2.47x even the four-times-slower case. NO NUMBER HERE IS
+  # A MEASUREMENT OF INVERTER B.
   #
   # DELIBERATELY NOT 180 - that is SAFETY-02's site hold and this is not it -
   # and not 150 or 200. 37 control cycles.