Reference tests: the variable rounds 77, 78 and 81 pass on SMLib d01d7b0; their known-failure marks come off

The new section builds their edges on the surfaces, so STEP export no longer finds them 4.6e-5
to 6.1e-5 off. develop's Reference tests (60393a718) report all three as passing while marked.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

tests/oracle/77-fillet-variable-linear/expected.json

@@ -1,6 +1,5 @@
 {
   "source": "analytic",
-  "known_failure": "SMLib variable fillet: STEP export finds edges 4.6e-5 to 6.1e-5 off their surfaces against 2e-5",
   "method": "box 40x30x20 with its top front edge (y=0, z=20) rounded at a radius running straight from 2 (x=0) to 5 (x=40): sections are quarter rounds, V = abc - (1 - pi/4) L (a^2 + ab + b^2) / 3; moments and the round's area integrated along the edge (tools/oracle-analytic/cases_fillets.py)",
   "body_count": 1,
   "volume_mm3": 23888.407044966672,

tests/oracle/78-fillet-variable-to-zero/expected.json

@@ -1,6 +1,5 @@
 {
   "source": "analytic",
-  "known_failure": "SMLib variable fillet: STEP export finds edges 4.6e-5 to 6.1e-5 off their surfaces against 2e-5",
   "method": "as 77 with the radius running from 4 (x=0) down to 0 (x=40), where the round closes to the corner: V = abc - (1 - pi/4) L a^2 / 3; the round is three-sided and the corner at x=40 stays",
   "body_count": 1,
   "volume_mm3": 23954.21827485812,

tests/oracle/81-fillet-spline-radius/expected.json

@@ -1,6 +1,5 @@
 {
   "source": "analytic",
-  "known_failure": "SMLib variable fillet: STEP export finds edges 4.6e-5 to 6.1e-5 off their surfaces against 2e-5",
   "method": "box 40x30x20 with its top front edge (y=0, z=20) rounded at a radius that is a clamped cubic B-spline in u = x/40, knots 0,0,0,0,0.25,0.6,1,1,1,1, control radii 2, 5, 1.5, 4, 1, 3 (r(0) = 2, r(40) = 3). Sections are quarter rounds; V, moments and areas integrated along the edge, exactly per knot span",
   "body_count": 1,
   "volume_mm3": 23933.601078437176,