refactor ListEntry used in collision

[psychocoderHPC]

Refactor changes introduced in #4745 to avoid void using void pointer as a place holder for the later used frame list.

[pordyna]

Still fails with:

Unhandled exception of type 'St13runtime_error' with message '/global/u2/p/pordyna/picongpu-setups/picongpu/thirdParty/cupla/alpaka/include/alpaka/event/EventUniformCudaHipRt.hpp(143) 'ret = TApi::eventQuery(event.getNativeHandle())' returned error  : 'cudaErrorIllegalAddress': 'an illegal memory access was encountered'!', terminating
terminate called after throwing an instance of 'std::runtime_error'
  what():  /global/u2/p/pordyna/picongpu-setups/picongpu/thirdParty/cupla/alpaka/include/alpaka/event/EventUniformCudaHipRt.hpp(190) 'TApi::eventSynchronize(event.getNativeHandle())' returned error  : 'cudaErrorIllegalAddress': 'an illegal memory access was encountered'!

[psychocoderHPC]

We will merge this PR before going back to the slow particle access in collision #4810.

We need to test in a physical setup if #4828 + #4825 is solving the Perlmutter issues @pordyna is observing.

[steindev]

Is it guaranteed that the division inSuperCellIdx / frameSize does not yield a floating point number? (A comment may state that)

[pordyna]

With both integers? I don't think so

[psychocoderHPC]

Both types a integral types, they can not end in floating point numbers, that are defined by the language.
A comment is not useful because it will increase the code size without extra new knowledge.

[ikbuibui]

Looks okay