Write simulation state on a background thread to avoid stalling the computation.

This should most beneficial for remote filesystems as used by the EVE and MSG-HPC clusters, for example the baseline simulation experiment introduced in !98 (merged) changes efficiency from

> seff 300893
Job ID: 300893
Cluster: eve
User/Group: reichold/oesa
State: COMPLETED (exit code 0)
Nodes: 1
Cores per node: 4
CPU Utilized: 03:05:02
CPU Efficiency: 99.98% of 03:05:04 core-walltime
Job Wall-clock time: 00:46:16
Memory Utilized: 44.18 MB
Memory Efficiency: 1.08% of 4.00 GB

to

> seff 300894
Job ID: 300894
Cluster: eve
User/Group: reichold/oesa
State: COMPLETED (exit code 0)
Nodes: 1
Cores per node: 4
CPU Utilized: 03:07:28
CPU Efficiency: 100.00% of 03:07:28 core-walltime
Job Wall-clock time: 00:46:52
Memory Utilized: 46.38 MB
Memory Efficiency: 1.13% of 4.00 GB