Merge T5 After Build

[GNiendorf]

Continuation of #236. This draft PR attempts to move the T5 merging kernel right after T5's are built to use the hit extensions from merged T5's in the downstream kernels (e.g. pT5 building, all duplicate cleaning kernels, etc.)

[GNiendorf]

run-ci: all
modifiers: gpu

[GNiendorf]

Something definitely broke when moving the commits over, let me see what happened.

[GNiendorf]

Looks like there is some issue just on GPU, need to look into it. With commit 27fdd75

[GNiendorf]

run-ci: all

[GNiendorf]

run-ci: all
modifiers: gpu

[github-actions]

The PR was built and ran successfully in standalone mode running on GPU. Here are some of the comparison plots.

The full set of validation and comparison plots can be found here.

Here is a timing comparison:

   Evt    Hits       MD       LS      T3       T5       pLS       pT5      pT3      TC       Reset    Event     Short             Rate
   avg     34.4      0.3      0.4      0.6      1.5      0.3      0.6      1.5      0.3      1.4      0.0      41.4       6.7+/-  2.9      41.4   explicit[s=1]
   avg      1.1      0.5      0.6      0.8      1.9      0.3      0.8      2.1      0.5      2.0      0.0      10.6       9.1+/-  3.2       5.4   explicit[s=2]
   avg      2.1      0.8      1.0      1.3      2.6      0.5      1.3      2.7      0.7      3.3      0.0      16.3      13.7+/-  4.1       4.2   explicit[s=4]
   avg      3.1      1.2      1.5      2.0      3.3      0.7      2.0      3.0      1.1      4.6      0.0      22.5      18.7+/-  5.1       3.9   explicit[s=6]
   avg      4.2      1.8      2.2      2.8      4.4      0.9      2.6      3.6      1.2      5.8      0.0      29.6      24.5+/-  5.4       3.8   explicit[s=8] (target branch)
   avg     34.5      0.3      0.4      0.6      1.5      0.3      0.6      1.5      0.3      1.4      0.0      41.5       6.7+/-  2.9      41.5   explicit[s=1]
   avg      1.1      0.5      0.6      0.8      1.9      0.3      0.8      2.1      0.5      2.0      0.0      10.5       9.1+/-  3.3       5.4   explicit[s=2]
   avg      1.9      0.7      1.0      1.3      2.7      0.5      1.4      2.6      0.7      3.2      0.0      16.0      13.5+/-  3.9       4.1   explicit[s=4]
   avg      3.0      1.2      1.5      1.9      3.6      0.7      2.0      3.0      1.0      4.6      0.0      22.5      18.8+/-  5.0       3.9   explicit[s=6]
   avg      3.8      1.9      2.1      2.7      4.3      0.9      2.8      3.5      1.3      5.8      0.0      29.2      24.5+/-  5.8       3.8   explicit[s=8] (this PR)

[github-actions]

The PR was built and ran successfully with CMSSW running on GPU. Here are some plots.

OOTB All Tracks

The full set of validation and comparison plots can be found here.