nx_shard_registry.nx

nx_shard_registry.nx -- the SSOT reader for the Nishi SHARDED ecosystem (CAP-SHARD-REGISTRY). Industry best-practice for managing many shards/tenants = ONE control plane over a REGISTRY of data planes (AWS Organizations' account tree, k8s fleet's cluster registry, a SaaS tenant catalog: "onboard a tenant = write a row to the registry"). This reads knowledge/hosting/shards.conf -- one row per shard = the JOIN the ecosystem lacked (domain x service x port x identity x roles x health, today scattered across ~8 confs). The control plane (nx_mgmt_api /api/shards) enumerates + ADDRESSES every shard by shard-id, and per-shard AUTH (realm) + per-shard RBAC (roles_path) become DATA not code -- which is exactly how one canonical realm scopes authority per shard WITHOUT the rejected dual-realm hack. Composes the shipped slk_* TSV parsers (nx_site_lock_lib). Add a shard = a row + reload, no recompile. row: shard_id <TAB> domain <TAB> service <TAB> port <TAB> realm <TAB> keys <TAB> store <TAB> roles_path <TAB> health_flow license_tier: ORIGINAL