test(clustering): fixture 5 multi_operator + c2/shift/composite refs

Three new reference clusterers in fixture_harness:

* c2_callback_clusterer — union-find on overlapping C2 callback
  sets across an attacker's sessions. Pass-clusterer for fixture 5
  where two operators with distinct tooling share a C2 endpoint as
  the campaign signal.

* shift_clusterer — deliberately-bad reference that buckets
  attackers by majority session-start hour into night/day/swing.
  Adversarial reference for fixture 5; proves operational schedule
  is NOT a campaign signal.

* composite_signals_clusterer — union-find combining (ja3, hassh)
  match OR overlapping C2 callback. Will serve as the pass-
  clusterer for fixture 6 (noise_floor) where multiple campaigns
  with heterogeneous signal types are scored together.

Also factored a small _union_find helper for the new clusterers
(existing time_window/credential_jaccard left untouched to avoid
mixing refactor with feature work).

Fixture 5 (multi_operator): one campaign, two operators with
distinct UKC roles. Actor A (broker, night shift): Delivery →
Exploitation → Persistence → C2. Actor B (post-ex, day shift):
Discovery → Lateral Movement → Collection → Exfiltration.
Distinct JA3/HASSH/ASN/IPs; shared C2 + payload hash.

Four tests: corpus shape (distinct fingerprints, shared C2,
disjoint shifts), pipeline pass via c2_callback_clusterer,
explicit harness sanity that fingerprint_clusterer cannot resolve
this fixture (documents which signal carries the campaign), and
adversarial shift_clusterer fragmentation.

Phase-handoff edges (the real load-bearing signal per the design
doc) wait for the production clusterer; this fixture will prove
they're needed when it ships.

tests/fixtures/campaigns/multi_operator.expected.yaml

@@ -0,0 +1,25 @@
+# Bounds for fixture 5 (multi_operator).
+#
+# Ground truth at campaign-level: 1 campaign of 2 observation rows
+# (one per DSL actor). A correct algorithm scores 1.0 across every
+# metric on this fixture.
+#
+# Completeness is the load-bearing metric: a clusterer that splits
+# the two operators by shift / by tooling / by ASN tanks
+# completeness (the one true class is split across two predicted
+# clusters). The adversarial shift_clusterer demonstrates this and
+# the bound below rejects it.
+#
+# Campaign-level fixture only — the two DSL actors model two
+# distinct identities (different tooling, different operators) by
+# design. See the YAML header for the modeling note.
+#
+# Bounds are loose at v1; tighten as the algorithm matures.
+adjusted_rand_index:
+  min: 0.85
+homogeneity:
+  min: 0.90
+completeness:
+  min: 0.80
+singleton_recall:
+  min: 0.95