Release v2.2.9

• new feature: two glycan resid numbering conventions are now available via sequence.glycans.numbering
  • narrow (default): glycan blocks start above the maximum protein resid of the parent chain; block size controlled by sequence.glycans.max_glycan_size (default 30)
  • wide: the root glycan monomer (directly bonded to the protein) gets the protein attachment resid plus sequence.glycans.wide_shift (default 3000); subsequent monomers increment by 1 in BFS order from the root (e.g. NAG on ASN 123 → resid 3123, next monomer → 3124)
• bugfix: --ncpus CLI option and namd.ncpus config key now correctly override NAMD PE count; the processor-info banner reflects the effective count at startup
• bugfix: under SLURM, NAMD (CPU/multicore mode) is now launched as numactl --localalloc namd3 +p N script.namd; both charmrun +p N and bare +p N bypass SLURM's CPU affinity mask, scattering Charm++ worker threads across all NUMA nodes and causing severe startup slowdown due to cross-NUMA memory access during bond-table construction; numactl --localalloc preserves Charm++'s thread placement while ensuring each thread's memory allocations land on its local NUMA node
• bugfix: glycan resid block allocation in the narrow convention now correctly starts above the parent chain's maximum protein resid rather than from 1, eliminating clashes between glycan and protein resids within the same chainID