Added Q criterion and vorticity calculation + VTK output to LBM fields

See merge request walberla/walberla!344

add user defined literal _r for real_t

See merge request walberla/walberla!132

Added new LBM tutorial for backward facing step setup

See merge request walberla/walberla!346

added getResidentMemorySize to get how much memory is currently used

See merge request walberla/walberla!351

introduce extra stage for merge requests

See merge request walberla/walberla!342

The new CI has four stages:
1. **pretest** jobs are run automatically as a first check
2. **test** jobs run at each commit but contain only a limited number of configurations
3. **merge_request** these jobs need to be run **manually** before merging a merge request
4. **benchmark** benchmarks are only run manually if needed

Changes in mesa_pd functionality for halfspaces

See merge request walberla/walberla!350

Fix typo: rebelance -> rebalance

See merge request walberla/walberla!339

Add missing BodyCast.h to GJKEPACollideFunctor.h

See merge request walberla/walberla!348

[FIX] Warning -Wmaybe-uninitialized

See merge request walberla/walberla!345

Bugfix: call cudaFree before resize of gpuVectors_

See merge request walberla/walberla!338

Add .editorconfig rule for Python to agree with flake settings

See merge request walberla/walberla!335

How could this have been working until now?
We should not call `cudaFree` on uninitialized memory.

We have to free the old pointers before resize. This also prevents
memory leakage in case the vector gets shrinked.

Apparently the previous code relied on the fact that `resize` somehow
initialized the memory to null pointers or that `cudaFree` just ignores
pointers it didn't allocate.

CustomMemoryBuffer: only memcpy when not `nullptr`

See merge request walberla/walberla!340

Replace some macros with C++11/14 standard functionality

See merge request walberla/walberla!341

add gcc 10 to CI

See merge request walberla/walberla!329

Suppress signedness warning in BlockExclusion.h

See merge request walberla/walberla!333

MESA-PD LB Benchmark Update

See merge request walberla/walberla!334

It is not needed because deprecations are compiler warnings and should not be ignored anyway.
Also, no need to export WARNING_DEPRECATED to the C++ code.

Our own wrappers was only used in one place anyway and did not all allow for specify a reason string.

Prepare compilation for C++17: std::any does not have `empty`

See merge request walberla/walberla!337

Silence recent Clang warning about unnecessary copy

See merge request walberla/walberla!336