WIP feat: add slurm cluster autodiscovery

[gregorweiss]

Adds mdfactory/performance/cluster.py — queries sinfo and sacctmgr to discover cluster resources automatically. This is the foundation for all planned HPC features (benchmarking, node packing, GPU sharing).

What it does

• discover_cluster() → returns partitions, node specs (CPUs, memory, GPUs), accounts, and QOS policies as frozen dataclasses
• select_partition(cluster, needs_gpu=True, min_cpus=64) → picks the best partition for given requirements
• Returns None gracefully on non-SLURM machines
• Results cached per session (topology doesn't change mid-run)

Files

• mdfactory/performance/__init__.py — new package
• mdfactory/performance/cluster.py — autodiscovery implementation
• mdfactory/tests/test_cluster.py — unit tests with mocked SLURM output

Test plan

• pytest mdfactory/tests/test_cluster.py -v (mocked, no cluster needed)
• Verify discover_cluster() returns None on laptop
• Verify populated ClusterInfo on a real SLURM node

[CLAassistant]

All committers have signed the CLA.