[lohedges]

Profiling checkpoints of a large replica exchange system in SOMD2 showed that most of the time was spent creating and destroying Sire objects for every molecule, with the GIL held, so replicas checkpointed in parallel mostly took turns.

• Release the GIL when updating a System or deleting its frames, and add a private _free() to System and SelectorMol to destroy large objects without it.
• Skip loading the current frame when saving DCD, TRR or XTC trajectories, since every frame is written from the trajectory itself.
• Add DCD.concatenate() to join DCD files without a topology, with frame times set from a timestep and start time.
• Fix DCD frames being read from the file twice.

Together with the matching SOMD2 changes, normal checkpoints of a 137,000-atom ABFE system are about 2.7 times faster, and the final checkpoint, which joins the trajectories, went from about 4.5 minutes to 13 seconds.

Profiled and written with assistance from Claude Code (Opus 5.5).

• I confirm that I have merged the latest version of devel into this branch before issuing this pull request (e.g. by running git pull origin devel): [y]
• I confirm that I have added a test for any new functionality in this pull request: [y]
• I confirm that I have added documentation (e.g. a new tutorial page or detailed guide) for any new functionality in this pull request: [y]
• I confirm that I have added a changelog entry to the changelog (we will add a link to this PR as part of the review): [y]
• I confirm that I have permission to release this code under the LGPLv3+ license: [y]