LBCPU: Fix rounding issue for grid size calculation

[RudolfWeeber]

Description of changes:
Replaces ceil by round in lbcpu n*agrid ~= box_l calculation + a test in which the old behavior resulted in a false positive

[fweik]

The Lattice class is a gift that just keeps on giving...

[codecov]

Codecov Report

Merging #3678 into python will increase coverage by 0%.
The diff coverage is 100%.

@@           Coverage Diff           @@
##           python   #3678    +/-   ##
=======================================
  Coverage      87%     87%            
=======================================
  Files         526     536    +10     
  Lines       21987   22758   +771     
=======================================
+ Hits        19301   20014   +713     
- Misses       2686    2744    +58     

Impacted Files	Coverage Δ
src/core/grid_based_algorithms/lattice.cpp	88% <100%> (+<1%)	⬆️
src/core/EspressoSystemInterface.hpp	48% <0%> (-19%)	⬇️
src/core/polymer.cpp	92% <0%> (-6%)	⬇️
src/core/particle_data.cpp	96% <0%> (-1%)	⬇️
src/script_interface/lbboundaries/LBBoundary.hpp	100% <0%> (ø)
...e/virtual_sites/VirtualSitesInertialessTracers.cpp	100% <0%> (ø)
.../grid_based_algorithms/lbboundaries/LBBoundary.hpp	100% <0%> (ø)
src/core/actor/Actor.hpp	100% <0%> (ø)
src/core/actor/DipolarDirectSum.cpp	100% <0%> (ø)
src/core/CudaHostAllocator.hpp	100% <0%> (ø)
... and 24 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ae36c6f...6986c2a. Read the comment docs.