nx_infill_test.nx

nx_infill_test.nx -- scan-line infill on a unit-mm CCW square at multiple densities, hand-computed expected counts and endpoints. Unit square (0,0) - (1mm, 1mm) in Q14: (0,0) - (16384, 16384). Line width = 0.4mm = 6554 Q14. Density 100% -> spacing = 6554 Q14. Scan lines at y = 0, 6554, 13108 (next 19662 > 16384 = out of bbox). 3 segments. Density 50% -> spacing = 13108 Q14. Scan lines at y = 0, 13108. 2 segments. Density 25% -> spacing = 26216 Q14. Scan line at y = 0 only. 1 segment. At each scan line, the square produces ONE crossing pair from x = 0 to x = 16384 (left and right vertical edges of the square). Horizontal edges of the square (top and bottom) are skipped by the straddle test, which is the right behaviour. Closed-form invariants: (a) 100% density horizontal scan -> exactly 3 segments. (b) 50% density horizontal scan -> exactly 2 segments. (c) 25% density horizontal scan -> exactly 1 segment. (d) Every emitted segment is horizontal (y1 == y2). (e) Every emitted segment spans full width (x1=0, x2=16384). (f) Vertical scan (vertical=1) at 50% density -> 2 vertical segments (x1==x2, span full height). (g) density_pct = 0 returns BAD_DENSITY verdict. (h) density_pct = 101 returns BAD_DENSITY verdict. (i) 3-vert polygon (triangle) at 100% density yields > 0 segments. expect_exit: 0 license_tier: ORIGINAL