Remove split CCL

[WeiyiVirtonomy]

The split cell linked list is replaced by a function called particle_for_split, where the cells are accessed by a stride of 3 in the indices.

TODO:
Add sequential version
Add unit test

[WeiyiVirtonomy]

I added the sequential version and unit tests

[WeiyiVirtonomy]

I had some misunderstanding about the method, will fix it tomorrow

[WeiyiVirtonomy]

I think this PR is ready for review now

[WeiyiVirtonomy]

@DongWuTUM The computational time of damping is too long. I ran the same simulation without and with damping, the previous one took 104s and the later one took 199s, nearly a 90% increment.
The one with block_2d loop + mod is close to the two 1D loop one. For some reason, the 2D parallel algorithm is much slower than the 1D parallel of split cell list vector

FYI, for test_3d_arch, the computational time on my machine is:
Old method: without damping 26.1719s, with damping 37.9359s, damping time 11.764s
New method: without damping 23,214s, with damping 47.5902s, damping time 24.3762s

The damping computational time of the new method is twice more than that of the old method

[Xiangyu-Hu]

@WeiyiVirtonomy I guess that there is something wrong.

[Xiangyu-Hu]

@WeiyiVirtonomy As the master has been updated, the conflicts must be resolved first for the merge. i also noticed that there are some function is base mesh have repeating implementation, please also clean these repeating if there is any.

[WeiyiVirtonomy]

@WeiyiVirtonomy As the master has been updated, the conflicts must be resolved first for the merge. i also noticed that there are some function is base mesh have repeating implementation, please also clean these repeating if there is any.

Merge conflict is now solved