Bug description

When writing checkpoint with particle tracking enabled, the particle counter is mistakenly saved with npart() (ln. 641-646 of framework/containers/particles_io.cpp)

out::WriteVariable<npart_t>(io,
                                writer,
                                fmt::format("s%d_counter", index()),
                                npart(),
                                domains_total,
                                domains_offset);

So the counter in the checkpoint is the number of active particles in the domain, which is not equal to the true counter. As a result multiple particles will have the same ID after restarting from checkpoint.

Suggested fix

Substituting npart() with counter().

The fix is tested on 2D shock setup with tracking enabled.