Not everything in CA  is excellent. See picture below:

  1.  CA reused hits. Where: In XY plane there are hits which where reused. You see that higher dencity near the boundaries of sectors
      It is probably result of track merging between different sectors.
      Merging is not excellent.
     Simulation:

 Real data:

     There is some difference. Amount of reused hits bigger for big Rxy, where dencity is the smallest and
     amount of reused hits should be minimal. Puzzle

      The same hits but in RZ coordinates. What is strange, that dencity of reused hits bigger with big Rxy where dencity of tracks the smallest
      No explanations!!! It must be the opposite.

    The same for real data:

   2.  In bug report # 3232. It is the case when removed 1st or last hit, and then 1st or last node parameters becames senseless.
       There are XY of such removed hits. You see that these begins & ends of tracks like to live near sectors boundaries.
       That means  merging is working not excellent

   3.  The same hits in RZ representation. Here we do not see anything interesting

  4. in CA bug #3233 we found that on some tracks hits are not properly ordered.
     These hits in XY plane: The impression that it mostly happens between inner and outer (120 cm)

Now the same for unsorted by padrow number:

There is no big difference.  

Now In RZ view

you cans see that at Z=0 (membrane) there is higher dencity

    And the same for the unsorted pads;

      So, unsorting is relatated for both, pads and Rxy.

   5. Now investigated the bug #3233 we found that not only disorder but also same hit encontered twice in the same track.
      There is XY view of these hits. Again they are concentrated in inner/outer boundaries.

   Same in RZ view:

      6. Now PercentOfReused(numberOfReused) for simulation

The same for real data:

  It is strange, but for real data result is better. Why?