Assigns a topic's ranges across the live members of a consumer group, so each range is consumed by exactly one member and the work spreads in parallel: the consumer-group counterpart of Kafka's partition assignment, over NorthGuard's dynamic ranges.

Reuses Malachi.Cluster.Placement.place/4 (rendezvous/HRW hashing): each range goes to its top-ranked member (replication_factor: 1). HRW is min-reshuffle, which is exactly the property a consumer group wants. strong stickiness: a member leaving moves only its own ranges (every surviving member keeps all of its ranges), and a member joining moves ranges only to it (existing members only ever lose ranges, never trade them). That matters because each reassignment costs the new owner a re-read from the group's committed position. Balance is statistical (HRW spreads evenly in expectation) rather than exactly even: the right trade for NorthGuard, whose hot topics accumulate many ranges as they split, so the law of large numbers keeps consumers well balanced while churn stays minimal. The result is deterministic (the same ranges and members yield the same assignment on every node), so a replicated or failed-over coordinator computes an identical one.