Shape Healing - initialize owires before ShapeAnalysis_FreeBounds's empty-input early return

[gsdali]

Fixes #1376.

Problem

ShapeAnalysis_FreeBounds's constructor (which runs SplitWires() internally) crashes with an
uncatchable SIGSEGV on some shapes containing multiple free-boundary components. Minimally
reproducible with just two disjoint planar faces in one compound — no shared or even nearby
geometry needed. Not a data-volume threshold: a shape with 150+ free-boundary loops can be fine
while a much smaller one crashes, depending only on whether any single component happens to close
with nothing left over.

Root cause

ShapeAnalysis_FreeBounds::SplitWire (the per-wire helper) finds each wire's closed sub-loops,
then hands whatever edges weren't consumed to ConnectEdgesToWires to chain into the "open"
result:

for (i = 1; i <= nbedges; i++)
  if (statuses.Value(i) != 2)
    edges->Append(sewd->Edge(i));
open = ShapeAnalysis_FreeBounds::ConnectEdgesToWires(edges, toler, shared);

When every edge of the wire was consumed into a closed loop, edges is a valid but empty
sequence. The call chain ConnectEdgesToWires → ConnectWiresToWires (3-arg) →
ConnectWiresToWires (4-arg, with a local occ::handle<...> owires;) → connectWiresToWiresImpl
starts with:

if (iwires.IsNull() || !iwires->Length())
{
  return;
}

For empty input this returns without ever assigning owires — every caller in the file starts
from a freshly-defaulted (null) handle, so the null propagates back through ConnectEdgesToWires's
return value into SplitWire's open output parameter, into
ShapeAnalysis_FreeBounds::SplitWires's per-wire loop:

SplitWire(TopoDS::Wire(wires->Value(i)), toler, shared, tmpclosed, tmpopen);
closed->Append(tmpclosed);
open->Append(tmpopen);          // tmpopen is null here

NCollection_HSequence::Append(const opencascade::handle<T>&) dereferences the null handle
(theOther->ChangeSequence()) to obtain a reference, which is then passed into
NCollection_Sequence::Append(NCollection_Sequence&), whose first statement reads
theSeq.IsEmpty() (== theSeq.mySize) through that invalid reference — SIGSEGV.

Fix

The one-line contract restoration connectWiresToWiresImpl's own non-empty path already follows a
few lines down (owires = new NCollection_HSequence<TopoDS_Shape>; before populating it): "no
wires to connect" should produce a valid empty result, not an untouched (and, for every caller
in this file, null-defaulted) out-parameter.

Reproducer / test

BRepBuilderAPI_MakeWire mw1;
mw1.Add(BRepBuilderAPI_MakeEdge(gp_Pnt(0,0,0), gp_Pnt(1,0,0)).Edge());
mw1.Add(BRepBuilderAPI_MakeEdge(gp_Pnt(1,0,0), gp_Pnt(1,1,0)).Edge());
mw1.Add(BRepBuilderAPI_MakeEdge(gp_Pnt(1,1,0), gp_Pnt(0,1,0)).Edge());
mw1.Add(BRepBuilderAPI_MakeEdge(gp_Pnt(0,1,0), gp_Pnt(0,0,0)).Edge());
TopoDS_Face faceA = BRepBuilderAPI_MakeFace(mw1.Wire()).Face();

BRepBuilderAPI_MakeWire mw2;                                    // far away, no shared geometry
mw2.Add(BRepBuilderAPI_MakeEdge(gp_Pnt(5,5,0), gp_Pnt(6,5,0)).Edge());
mw2.Add(BRepBuilderAPI_MakeEdge(gp_Pnt(6,5,0), gp_Pnt(6,6,0)).Edge());
mw2.Add(BRepBuilderAPI_MakeEdge(gp_Pnt(6,6,0), gp_Pnt(5,6,0)).Edge());
mw2.Add(BRepBuilderAPI_MakeEdge(gp_Pnt(5,6,0), gp_Pnt(5,5,0)).Edge());
TopoDS_Face faceB = BRepBuilderAPI_MakeFace(mw2.Wire()).Face();

BRep_Builder B; TopoDS_Compound comp; B.MakeCompound(comp);
B.Add(comp, faceA); B.Add(comp, faceB);

ShapeAnalysis_FreeBounds safb(comp, 0.01, true, true);  // crashes on master

Crashes on current master; fixed here. Added GTest
ShapeAnalysis_FreeBoundsTest.TwoDisjointFacesNoCrash (plus a single-face baseline).

Validation

Isolated with AddressSanitizer (an ASan-instrumented macOS-arm64 build of ModelingAlgorithms +
ModelingData + FoundationClasses, RelWithDebInfo, MMGT_OPT=0 at runtime so allocations route
through malloc/free): the two-disjoint-face repro crashes 100% of the time on stock master with an
identical signature at both -O2 and -O0 (same function, same NCollection_Sequence::Append
call, same 0xfffffffffffffff8 fault address), and returns the correct 2 closed, 0 open after
this patch. On a real-world 150-face analytic-BRep fixture (sewn with BRepBuilderAPI_Sewing):
tol=0.05 gives 152 closed / 0 open byte-identical before and after this patch (confirming no
behavior change on the working path); tol=0.10 crashes on stock master and returns
144 closed / 0 open after.

Reported downstream and isolated at SecondMouseAU/OCCTSwift#310.