Perform more repetitions (of increment) for RTC concurrency testing

[ptheywood]

RTC runtime was too fast for reliable timing, so a longer running kernel was required.

Closes #745

[ptheywood]

@Robadob - could you test this when you have time to see if this has resolved the failure for you (windows, 1080), by running the following a number of times and seeinf if it ever fails?
First run will be a while as it RTC's 4 kernels, subsequent ones should be 80 to 200ms depending on how long the cuda context takes to initialsie on windows + a little bit of time.

./bin/Release/tests --gtest_filter="TestCUDASimulationConcurrency.RTCLayerConcurrency"

(I'll re-run CI tomorrow when github's hopefully more stable)

I've ran it ~ 25 times successfully on my 3060ti on windows (current master failed once for me, but only when running the full TestCudaSimulationConcurreny suite, not just the RTC test so this may not be 100%)

[Robadob]

I'll try and remember to do it tomorrow,

[Robadob]

Just ran the version that failed yesterday again, with less other stuff using GPU, and it passed.

Rebuilt it with this branch and it also passed.

[ptheywood]

I'll assume that it's good for now then, although we may need to adjust this further in the future