Enhance fillet handling and add validation for geometry integrity

[angelobartolome]

This pull request addresses issue #1177 by improving the robustness of fillet and chamfer operations when the radius or distance is large enough to entirely consume adjacent faces. The changes ensure that degenerate or invalid faces are detected and removed, allowing fillets and chamfers to meet cleanly without leaving behind problematic geometry. Additional regression tests are added to verify correct behavior in these edge cases.

Robustness improvements for fillet/chamfer operations:

• Added the function ChFi3d_RemoveConsumedFaces in ChFi3d_Builder.cxx to detect and remove faces that have been fully consumed (collapsed to a negligible area) by fillet or chamfer operations, and to re-stitch the remaining faces into a valid solid or shell. This function is now called during the result assembly in ChFi3d_Builder::Compute(). [1] [2]
• Introduced helper functions ChFi3d_IsConsumedFace and ChFi3d_IsPlanar to identify faces that should be removed.

Intersection and hatching logic improvements:

• Updated intersection checks in ChFi3d_Builder_0.cxx and ChFi3d_Builder_C1.cxx to distinguish between transversal (invalid) and tangential (valid) fillet/chamfer intersections, allowing tangentially meeting fillets/chamfers to proceed without error. [1] [2] [3]
• Modified hatching logic in ChFi3d_Builder_SpKP.cxx to retain points and segments, enabling valid domains to be created even when a fillet's tangency line coincides with a face boundary.

Regression tests for consumed faces and edge cases:

• Added comprehensive regression tests in BRepFilletAPI_MakeFillet_Test.cxx and BRepFilletAPI_MakeChamfer_Test.cxx to verify that fillet and chamfer operations succeed and produce valid solids when faces are fully consumed, including cases where fillets/chamfers meet on opposing edges or remove entire faces. [1] [2]

Dependency and include updates:

• Added necessary includes for geometry, bounding, and sewing operations in ChFi3d_Builder.cxx.

Discussion
Issue #172

[dpasukhi]

Amazing job! If you will need anything from me - let me know.
I'm open to the call/emails to discuss the patch.

In case when you complete the implementation, I would be happy to review and prepare to the OCCT rules and maybe fix something (usually the final steps I do by myself, they are mostly specific for the project).
I also will run full test round on your patch when you will ask me (even on the intermediate version).

[angelobartolome]

Thanks @dpasukhi !

I think this PR is good to go as is right now, it will solve a bunch of use cases already.

I'm exploring extending this to allow for overflowing fillets/chamfers, but that is harder and bigger task, I'll probably do it in another PR or discussion if I succeed.

I see two failing checks in the pipelines, I think we need to check those before continuing

[dpasukhi]

I will analyze regressions and also will have a full circle of testing (on our side we have x2+ tests)
It will take some time, but assume up to that Sunday evening I will have some patch/report to you. I will push to your branch In case of some patches.
(sorry for delay, now I'm a little busy with hot patch for 8.0.0 :(

[dpasukhi]

17 FAILED Test failed (regression)
3 IMPROVEMENT Possible improvement

only 2 real regressions, but it is anyway were before, all the other FAILED - improvements.

There is a issue with performance, because you add BRepCheck + Sewing on each operation, which is not cheap.
I will clean up the code from most of the comments and some other generic clean up and code improvements.
+9-100% on some operations (it depends)

But it will take me some time to analyze the possibility to avoid BRepCheck and Sewing, because it is too expensive just to be and we usually avoid extra complex operations on the middle of others.

Also
Don't feel it right ;) I will need some time to somehow improve that part

    try
    {
      OCC_CATCH_SIGNALS
      BRepCheck_Analyzer anAnalyzer(theShape);
      isShapeValid = anAnalyzer.IsValid();
    }
    catch (const Standard_Failure&)
    {
      isShapeValid = true;
    }

[angelobartolome]

Sounds good, and I agree with the performance issue, let me know if I can help in anyway!

[luzpaz]

@dpasukhi have you had more thoughts on how to workaround the performance and methodology issues you raised in #1293 (comment) ?

[dpasukhi]

@dpasukhi have you had more thoughts on how to workaround the performance and methodology issues you raised in #1293 (comment) ?

There are no needs to thoughts. The fix will require the deep debug and proper writing the cases.
The problem is not performance, but the complexity of the solution.

That part still pending when I will have at least a few day to debug.
But using BRepCheck+Sewing inside the other operatings unfortunately is not acceptable to merge.

[jhaand]

Any update on this?

[dpasukhi]

I started to work on it. It will take about 1+-1 week for me approximately (I can't fully allocate the time on R&D :(
The task is planned to dev release or 8.1.