_fp_try_autofinish_on_drain guarded on _fp_has_real_incoming() — the WRONG
direction. The first stage (e.g. Racking) is fed by the qty_at_step seed,
not an incoming move, so it never auto-finished when all its parts were
sent forward (operator sent everything out of Racking, step stayed
in_progress at qty 0). Now guards on a real OUTGOING move (parts left),
which covers the seeded first stage.
Still best-effort + gated: button_finish runs the required-input / sign-off
gates, so a step with an unrecorded required input (Racking's "Count the
Parts") won't auto-finish — it stays in_progress for a manual finish after
the input is recorded. Verified on entech.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
7fcf38ca82