tests: fix the EN-115 ownership probes so proof B reads 118/118 (CR-139) - #446

… and fix the S notes (CR-139)

engine/margin_required_explicit_overflow pinned a failure
(engine_command_value_invalid, margin.required) for a finite explicit
quantity whose placement requirement overflows. L6's 10-14 ruling (f2a455b,
"row-7 is re-scoped") keeps that case: the order is refused at signal time
with cancellation and callback continuation, and the typed fault stays for
nonfinite placement inputs, nonfinite equity and declaration-derived
requirements. The engine already matches it (pine_adapter.cpp:7365-7378, and
tests/test_value_invalid_sizing_margin_source.cpp:196-197 and :232-262), so
the expectation becomes completed / null / {} / min_trades 0 with label
control and cites the ruling and the native test, not an observation. The
six *_unconstructible notes said the schema records an empty code; they now
state the PF-E0036 refusal the expectation carries (README:92-93).

Co-Authored-By: Claude Sonnet 5.5 <[email protected]>

…he rows they pin (CR-139)

engine/margin_maintenance_overflow entered 1e-10 units, which never became a
position, so the maintenance producer (which returns while the book is flat)
was never reached and the run completed with no trade. The entry is now one
unit; on a box the run fails with engine_command_value_invalid
(margin.maintenance, nonfinite), the pinned result.

engine/settlement_pnl_overflow failed one stage early: the percent fee quote
overflows to NaN before the settlement money predicate runs. A percent
commission of 0 cannot avoid that for a single trade, so the probe now uses a
price-free cash_per_contract commission of 0 and two lots that are closed
together, which takes the staged settlement and fails with engine_invariant at
the money predicate, the pinned result.

Both expectations are unchanged (they come from the ruled rows, not from an
observation); only probe.pine and NOTES.md change.

Co-Authored-By: Claude Sonnet 5.5 <[email protected]>

…CR-139)

Notes and one citation only; no expectation value changes. The row-7 notes
cite the main-line commit 868aa65 beside f2a455b, point the refusal at
the slipped-signal lines 7391-7421 for the configuration the probe uses, and
say that min_trades 0 is a floor so the drop is witnessed natively, not by
the probe. The six unconstructible notes no longer say the boundary row is
red on main (the host admits the declaration before copying it). The
margin_maintenance_overflow notes state the real extent of the 1x-long gap
(the replacement margin call returns early on any of: no lot grid, point
value other than 1, commission, slippage, an FX schedule, risk rules) and the
measured scratch results. The settlement_pnl_overflow notes give the code
argument for the staged path; the earlier commit message states that path as
a fact, but it is the design: the driver cannot tell the staged path from the
fused one.

Co-Authored-By: Claude Sonnet 5.5 <[email protected]>

…375795a) into the notes fix

Brings engine main (the #445 test sharding) and the merge commit made on the
pull request, and keeps the notes-only commit c4b0bd0 on top, so the branch
head is a fast-forward of the pull request head.

Co-Authored-By: Claude Sonnet 5.5 <[email protected]>