Add CPU and CUDA benchmark for toy detector reconstruction

[beomki-yeo]

This is for benchmarking the general speed of the track reconstruction in the toy geometry. (There will be a follow-up PR for monitoring)

Because the geometry and simulation data is not very forward compatible, the simulation is also done in the benchmark suite before the reconstruction. Of course, the simulation time is not included in the timing measurement.

Unfortunately, the simulation does not support the digitization, the performance of clusterization is not included in this benchmark.

Following is the output of CPU (+OpenMP) and CUDA benchmark with the single precision on my laptop:

CPU

[beomki@device-11 traccc_build]$ ./bin/traccc_benchmark_cpu 
2024-08-14T10:31:38+02:00
Running ./bin/traccc_benchmark_cpu
Run on (12 X 4367.4 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x6)
  L1 Instruction 32 KiB (x6)
  L2 Unified 256 KiB (x6)
  L3 Unified 12288 KiB (x1)
Load Average: 3.16, 3.19, 2.06
WARNING: No entries in volume finder

Detector check: OK
-------------------------------------------------------------------
Benchmark                         Time             CPU   Iterations
-------------------------------------------------------------------
ToyDetectorBenchmark/CPU 1.4495e+10 ns   1.3887e+10 ns            1

CUDA

[beomki@device-11 traccc_build]$ ./bin/traccc_benchmark_cuda
2024-08-14T10:33:49+02:00
Running ./bin/traccc_benchmark_cuda
Run on (12 X 4344.84 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x6)
  L1 Instruction 32 KiB (x6)
  L2 Unified 256 KiB (x6)
  L3 Unified 12288 KiB (x1)
Load Average: 2.94, 3.16, 2.20
WARNING: No entries in volume finder

Detector check: OK
--------------------------------------------------------------------
Benchmark                          Time             CPU   Iterations
--------------------------------------------------------------------
ToyDetectorBenchmark/CUDA 4734637722 ns   4719869245 ns            1

[krasznaa]

I'm absolutely on board with the addition of these benchmarks. Just have a number of technical comments...

[krasznaa]

Once you rebase it, it should be good to go. 👍