fix: any direction for helical track linearizer [backport #1363 to de…

…velop/v19.x] (#1377)

Backport 8639c37 from #1363.
---
Previously the `HelicalTrackLinearizer` assumed that the vertex we are fitting has to be reachable by back propagation. This is not always the case. In this PR we try to estimate the direction in which we will find the vertex and use it for the propagation.

@asalzburger mentioned that it would be good to back propagate from the first measurement of the particle on the track as this will always be "ahead" of the vertex

cc @paulgessinger

Core/include/Acts/Vertexing/HelicalTrackLinearizer.ipp

@@ -19,17 +19,21 @@ Acts::Result<Acts::LinearizedTrack> Acts::
   const std::shared_ptr<PerigeeSurface> perigeeSurface =
       Surface::makeShared<PerigeeSurface>(linPointPos);
 
+  auto intersection = perigeeSurface->intersect(gctx, params.position(gctx),
+                                                params.unitDirection(), false);
+
   // Create propagator options
   auto logger = getDefaultLogger("HelTrkLinProp", Logging::INFO);
   propagator_options_t pOptions(gctx, mctx, LoggerWrapper{*logger});
-  pOptions.direction = NavigationDirection::Backward;
+  pOptions.direction = intersection.intersection.pathLength >= 0
+                           ? NavigationDirection::Forward
+                           : NavigationDirection::Backward;
 
   const BoundTrackParameters* endParams = nullptr;
   // Do the propagation to linPointPos
   auto result = m_cfg.propagator->propagate(params, *perigeeSurface, pOptions);
   if (result.ok()) {
     endParams = (*result).endParameters.get();
-
   } else {
     return result.error();
   }