introduce shift in as built model for the QL3 modules

The origin of the local coordinate system in the sTGC QL3 modules is not at the center of the active area but at the cutout. This introduced a correlation between the scale and the offset parameters in the stgc internal alignment fit, therefore the origin was moved in the fit. This MR introduces this shift when applying the as-built parameters.

cc @itrigger

MuonSpectrometer/MuonAlignment/MuonAlignmentData/src/sTGCAsBuiltData2.cxx

@@ -19,7 +19,12 @@ Amg::Vector2D sTGCAsBuiltData2::correctPosition(const Identifier& channelId, con
     Amg::Vector2D correctedPos = pos;
     Parameters pars = par_itr->second;
     constexpr double convScale = 1.e-3; // parameters are stored in um and mrad therefore dividing by 1000
-    correctedPos.x() = pos.x() + (pars.offset * convScale  + pars.rotation *convScale * pos.y() + pars.scale*convScale*pos.x() + pars.nonPara * convScale * convScale * pos.x()*pos.y()); 
+    
+    // For the QL3 modules described by the legacy (Run 3) geometry EDM the origin of the local frame is shifted with respect to the actual center of the gap.
+    // Therefore a shift is introduced which stabilizes the as built fit which must also be applied here since the parameters are expressed using the actual center of the gap.
+    float shift = (std::fabs(m_idHelperSvc->stationEta(channelId)) == 3 && m_idHelperSvc->stationNameString(channelId) == "STL" ?  24.74 : 0.0);
+
+    correctedPos.x() = pos.x() +  (pars.offset * convScale  + pars.rotation *convScale * pos.y() + pars.scale*convScale*(pos.x() + shift ) + pars.nonPara * convScale * convScale * (pos.x() + shift  )*pos.y()); 
     return correctedPos;
 }
 
@@ -34,4 +39,4 @@ StatusCode sTGCAsBuiltData2::setParameters(const Identifier& channelId, const Pa
     }
     return StatusCode::SUCCESS;
 }
-    
\ No newline at end of file
+