Robustness goldens: record seed 13 after #1164 (#1268)

#1164 joins a body that sits flush on a rounded rim into it. In seed 13 the
rectangular pattern's second box (x 5..7.5) sits flush on the half-ring's top
face with a 27 mm2 contact patch. It used to stay a separate shell (one body,
three shells); now it joins the ring (area -54 mm2, twice the patch). The
fillet edge and hole position the generator picks next come from the new
geometry, so the end state differs: 3 bodies, 2647.95 mm3 (was 4, 2300.91).
Every step stays a sound solid, and no other seed moves.

Co-authored-by: Claude Opus 5.5 <[email protected]>

tests/robustness/goldens.json

@@ -14,7 +14,7 @@
   "10": {"area":2786.620097,"bodies":1,"edges":44,"error":null,"faces":18,"outcome":"ok","rejected":1,"shells":1,"steps":["revolved section","cut box","chamfer","hole"],"unsupported":0,"volume":6668.9804},
   "11": {"area":7858.300481,"bodies":1,"edges":33,"error":null,"faces":13,"outcome":"ok","rejected":0,"shells":1,"steps":["extruded profile","hole","rectangular pattern","fillet"],"unsupported":0,"volume":34605.048259},
   "12": {"area":4668.664174,"bodies":2,"edges":59,"error":null,"faces":25,"outcome":"ok","rejected":0,"shells":2,"steps":["extruded profile","chamfer","join cylinder","join box","cut cylinder","circular pattern"],"unsupported":3,"volume":11874.471611},
-  "13": {"area":1857.434638,"bodies":4,"edges":75,"error":null,"faces":34,"outcome":"ok","rejected":1,"shells":4,"steps":["revolved section","cut cylinder","join box","rectangular pattern","cut cylinder","cut box","fillet","hole"],"unsupported":0,"volume":2300.91061},
+  "13": {"area":1967.991667,"bodies":3,"edges":85,"error":null,"faces":37,"outcome":"ok","rejected":1,"shells":3,"steps":["revolved section","cut cylinder","join box","rectangular pattern","cut cylinder","cut box","fillet","hole"],"unsupported":0,"volume":2647.948638},
   "14": {"area":10553.267409,"bodies":1,"edges":43,"error":null,"faces":20,"outcome":"ok","rejected":0,"shells":1,"steps":["extruded profile","cut cylinder","shell","fillet"],"unsupported":0,"volume":10369.008764},
   "15": {"area":1974.784048,"bodies":1,"edges":40,"error":"boolean Cut: the operation failed: no result; the operation ","faces":18,"outcome":"failed","rejected":1,"shells":2,"steps":["revolved section","fillet","cut box","rectangular pattern","cut sphere","cut cylinder","cut sphere"],"unsupported":0,"volume":4781.224576},
   "16": {"area":8872.873884,"bodies":3,"edges":49,"error":null,"faces":23,"outcome":"ok","rejected":0,"shells":3,"steps":["box","join cylinder","circular pattern","hole"],"unsupported":0,"volume":44953.403148},
@@ -114,6 +114,7 @@
   {"reason":"hole tangent inside a stepped boss: touching sews split zero-thickness curved walls, so a pattern of holes spaced by their diameter (touching) in a stepped body cuts exactly instead of by a tool nudge: 42 builds more faces (the walls split where the holes touch), 51 picks other edges afterwards","seeds":[42,51]},
   {"reason":"corner blends on curved faces (#854): seed 33's last fillet (its second body) is built differently; its display-plan and measured volumes now agree (4442.73 mm³; on main they read 4439.42 and 4443.51)","seeds":[33]},
   {"reason":"#952: a hole placed on a face drills into the material; seed 81's hole moves its volume by 0.1 mm³ (area by 0.04 mm²)","seeds":[81]},
-  {"reason":"Cuts join the straight edges they used to leave split along untouched walls (#1049). Replaying each changed seed's new commands on the build before and after gives the same volume, area and faces at every step, and the same refusals (seeds 2 and 95 end on a cut the old build refuses too), except: seed 94's final join was 1.7 mm³ too large before (the box adds 875 − 2.5 × 97.967 = 630.08 mm³ to 5270.48; now 5900.56, before 5902.27), and seed 71's hole differs only in the coarse measuring mesh (identical at 2e-4 mm). Every other volume change is the generator picking other edges by index for fillets and chamfers, some of which are now accepted; seed 71's join now builds.","seeds":[2,4,7,12,13,17,25,31,38,42,45,51,52,56,65,67,69,71,72,74,82,93,94,95]}
+  {"reason":"Cuts join the straight edges they used to leave split along untouched walls (#1049). Replaying each changed seed's new commands on the build before and after gives the same volume, area and faces at every step, and the same refusals (seeds 2 and 95 end on a cut the old build refuses too), except: seed 94's final join was 1.7 mm³ too large before (the box adds 875 − 2.5 × 97.967 = 630.08 mm³ to 5270.48; now 5900.56, before 5902.27), and seed 71's hole differs only in the coarse measuring mesh (identical at 2e-4 mm). Every other volume change is the generator picking other edges by index for fillets and chamfers, some of which are now accepted; seed 71's join now builds.","seeds":[2,4,7,12,13,17,25,31,38,42,45,51,52,56,65,67,69,71,72,74,82,93,94,95]},
+  {"reason":"#1164: a pattern copy of a box sitting flush on the half-ring's top face (a 27 mm2 contact patch) now joins the ring instead of staying a separate shell, so the later random steps pick different edges and positions","seeds":[13]}
  ]
 }