D113 — failure event explains what rejected the path

Decision

A failure event says what rejected the path (issue #8): the attempt's track width, clearance and the narrowest legal width, the bottleneck of the best way from the closest approach (gap_nm with routed copper counted, gap_fixed_nm with only fixed copper), the path that was tried, and a verdict: contested (fixed copper leaves room, routed copper closes the way), rule (a narrower legal track or the board minimum clearance passes), placement (nothing legal passes), unknown when the evidence is incomplete. The gap is a widest-path search over the router's own obstacle tests (doc 13, "The failure verdict"), measured at a connection's 1st, 2nd, 4th, 8th ... failure, with a deterministic budget, and it counts no work

Alternatives

Measuring the free distance at the blocked lattice neighbour of the closest approach (it measures the approach to a wall, not the gap in it: a 0.3 mm gap with 0.2 mm clearance read 0.14 mm free); re-running the router's own search at other widths (it counts expansions against the work budget and would make a viewer change the run)

Reason

The label has to be a number the lattice would agree with, and a viewer must not change what is routed. The unknown verdict keeps a failed search alone from sending someone to a footprint move or a clearance change (the comment on the issue)