[STF] Ensure we generate CUDA graphs which always have the same topology

[caugonnet]

Some algorithms may reorder the edges in a CUDA graph, resulting in different topologies for the same code, and therefore prevent to update executable graphs.

Description

closes

Checklist

• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[caugonnet]

/ok to test f9fd8a9

[caugonnet]

/ok to test f9fd8a9

[github-actions]

🟩 CI finished in 41m 29s: Pass: 100%/26 | Total: 6h 14m | Avg: 14m 23s | Max: 28m 52s | Hits: 67%/14642

• 🟩 cudax: Pass: 100%/26 | Total: 6h 14m | Avg: 14m 23s | Max: 28m 52s | Hits: 67%/14642

  🟩 cpu
    🟩 amd64              Pass: 100%/22  | Total:  5h 22m | Avg: 14m 39s | Max: 28m 52s | Hits:  68%/12298 
    🟩 arm64              Pass: 100%/4   | Total: 51m 43s | Avg: 12m 55s | Max: 14m 12s | Hits:  61%/2344  
  🟩 ctk
    🟩 12.0               Pass: 100%/3   | Total: 39m 48s | Avg: 13m 16s | Max: 14m 27s | Hits:  68%/1463  
    🟩 12.8               Pass: 100%/23  | Total:  5h 34m | Avg: 14m 32s | Max: 28m 52s | Hits:  66%/13179 
  🟩 cudacxx
    🟩 nvcc12.0           Pass: 100%/3   | Total: 39m 48s | Avg: 13m 16s | Max: 14m 27s | Hits:  68%/1463  
    🟩 nvcc12.8           Pass: 100%/23  | Total:  5h 34m | Avg: 14m 32s | Max: 28m 52s | Hits:  66%/13179 
  🟩 cudacxx_family
    🟩 nvcc               Pass: 100%/26  | Total:  6h 14m | Avg: 14m 23s | Max: 28m 52s | Hits:  67%/14642 
  🟩 cxx
    🟩 Clang14            Pass: 100%/2   | Total: 26m 18s | Avg: 13m 09s | Max: 13m 37s | Hits:  61%/1176  
    🟩 Clang15            Pass: 100%/1   | Total: 14m 26s | Avg: 14m 26s | Max: 14m 26s | Hits:  61%/586   
    🟩 Clang16            Pass: 100%/1   | Total: 15m 20s | Avg: 15m 20s | Max: 15m 20s | Hits:  61%/586   
    🟩 Clang17            Pass: 100%/1   | Total: 15m 21s | Avg: 15m 21s | Max: 15m 21s | Hits:  61%/586   
    🟩 Clang18            Pass: 100%/1   | Total: 14m 40s | Avg: 14m 40s | Max: 14m 40s | Hits:  61%/586   
    🟩 Clang19            Pass: 100%/4   | Total: 46m 48s | Avg: 11m 42s | Max: 14m 19s | Hits:  71%/2344  
    🟩 GCC10              Pass: 100%/2   | Total: 29m 27s | Avg: 14m 43s | Max: 15m 00s | Hits:  61%/1176  
    🟩 GCC11              Pass: 100%/1   | Total: 16m 07s | Avg: 16m 07s | Max: 16m 07s | Hits:  61%/586   
    🟩 GCC12              Pass: 100%/1   | Total: 17m 15s | Avg: 17m 15s | Max: 17m 15s | Hits:  61%/586   
    🟩 GCC13              Pass: 100%/8   | Total:  1h 34m | Avg: 11m 52s | Max: 15m 27s | Hits:  70%/4688  
    🟩 MSVC14.39          Pass: 100%/1   | Total: 12m 40s | Avg: 12m 40s | Max: 12m 40s | Hits:  95%/287   
    🟩 MSVC14.42          Pass: 100%/1   | Total: 13m 41s | Avg: 13m 41s | Max: 13m 41s | Hits:  95%/287   
    🟩 NVHPC25.3          Pass: 100%/2   | Total: 57m 16s | Avg: 28m 38s | Max: 28m 52s | Hits:  58%/1168  
  🟩 cxx_family
    🟩 Clang              Pass: 100%/10  | Total:  2h 12m | Avg: 13m 17s | Max: 15m 21s | Hits:  65%/5864  
    🟩 GCC                Pass: 100%/12  | Total:  2h 37m | Avg: 13m 08s | Max: 17m 15s | Hits:  67%/7036  
    🟩 MSVC               Pass: 100%/2   | Total: 26m 21s | Avg: 13m 10s | Max: 13m 41s | Hits:  95%/574   
    🟩 NVHPC              Pass: 100%/2   | Total: 57m 16s | Avg: 28m 38s | Max: 28m 52s | Hits:  58%/1168  
  🟩 gpu
    🟩 h100               Pass: 100%/2   | Total: 19m 05s | Avg:  9m 32s | Max: 11m 19s | Hits:  80%/1172  
    🟩 rtx2080            Pass: 100%/24  | Total:  5h 55m | Avg: 14m 47s | Max: 28m 52s | Hits:  65%/13470 
  🟩 jobs
    🟩 Build              Pass: 100%/23  | Total:  5h 48m | Avg: 15m 10s | Max: 28m 52s | Hits:  62%/12884 
    🟩 Test               Pass: 100%/3   | Total: 25m 24s | Avg:  8m 28s | Max:  9m 49s | Hits:  99%/1758  
  🟩 sm
    🟩 90                 Pass: 100%/3   | Total: 30m 22s | Avg: 10m 07s | Max: 11m 19s | Hits:  73%/1758  
    🟩 90a                Pass: 100%/1   | Total: 12m 15s | Avg: 12m 15s | Max: 12m 15s | Hits:  61%/586   
  🟩 std
    🟩 17                 Pass: 100%/4   | Total:  1h 04m | Avg: 16m 06s | Max: 28m 24s | Hits:  60%/2342  
    🟩 20                 Pass: 100%/22  | Total:  5h 09m | Avg: 14m 04s | Max: 28m 52s | Hits:  68%/12300 
  

👃 Inspect Changes

Modifications in project?

	Project
	CCCL Infrastructure
	libcu++
	CUB
	Thrust
+/-	CUDA Experimental
	stdpar
	python
	CCCL C Parallel Library
	Catch2Helper

Modifications in project or dependencies?

	Project
	CCCL Infrastructure
	libcu++
	CUB
	Thrust
+/-	CUDA Experimental
	stdpar
	python
	CCCL C Parallel Library
	Catch2Helper

🏃‍ Runner counts (total jobs: 26)

#	Runner
17	linux-amd64-cpu16
4	linux-arm64-cpu16
2	windows-amd64-cpu16
2	linux-amd64-gpu-rtx2080-latest-1
1	linux-amd64-gpu-h100-latest-1

[miscco]

Nitpick: we should reserve the space

Suggested change

	for (cudaGraphNode_t node : nodes)
	result.reserve(nodes.size());
	for (cudaGraphNode_t node : nodes)