[mini] clean FillFromDeviceBuffer

[SeverinDiederichs]

This PR cleans FillFromDeviceBuffer kernel:

When looping over the particles to be copied back, the index starts at nAlreadyTransfered and was going to nAlreadyTransferred + maxFromDeviceBuffer. Via an if statement, out of bounds access was prevented.
This PR proposes to directly fix the range correctly in the loop, enabling to remove the out of bounds check later. Thus, this is some minor cleaning PR

Found via code analysis by @dkonst13.

[phsft-bot]

Can one of the admins verify this patch?

[SeverinDiederichs]

CI is failing because the shipped VecGeom misses some files like VecGeomConfigVersion.cmake. Under investigation

[JuanGonzalezCaminero]

I think this is not an issue because the index is checked before accessing the buffer:
https://github.com/apt-sim/AdePT/blob/master/include/AdePT/core/AsyncAdePTTransport.cuh#L278

Otherwise it would be crashing

[SeverinDiederichs]

Indeed! As discussed offline, with the correct loop bounds, the if statement can then be removed. So this is rather a cleaning PR :)

[SeverinDiederichs]

Failing just the known unstable reproducibility test in sync AdePT, therefore I am merging it now