chore(run-ops split): strip plan-enumeration labels from replication comments/test names

Remove the internal plan-enumeration labels from runs-replication
comments and test names, keeping the behavioral descriptions intact.
Comment/label hygiene only; no product logic or test behavior changed.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: d-cs

apps/webapp/app/services/runsReplicationInstance.server.ts

@@ -151,7 +151,7 @@ function initializeRunsReplicationInstance() {
   ]);
 
   if (env.RUN_REPLICATION_ENABLED === "1") {
-    // Shape B (construct-after-gate): resolve the async split gate ONCE at boot, and
+    // Construct-after-gate: resolve the async split gate ONCE at boot, and
     // when both sources are enabled rebuild `service` with sources[] before starting.
     // The legacy-only instance above is never started in the dual path (no slot/lock
     // taken). runsReplicationService.server.ts is untouched. The create route also calls

apps/webapp/test/runsReplicationService.part8.test.ts

@@ -194,7 +194,7 @@ describe("RunsReplicationService (part 8/8) - dual-source dedup", () => {
     }
   );
 
-  // Case A - reverse-order independence (Step 4.1).
+  // Case A - reverse-order independence.
   // Same run in both sources, but we flush the NEW (gen-1) snapshot FIRST, then the LEGACY
   // (gen-0) snapshot. The gen-1 winner must survive regardless of arrival order - the collapse
   // is FINAL-time and ordered by _version (composed origin generation), not by arrival time.
@@ -362,7 +362,7 @@ describe("RunsReplicationService (part 8/8) - dual-source dedup", () => {
     }
   );
 
-  // Case B - per-source independence / no cross-contamination (Step 4.2).
+  // Case B - per-source independence / no cross-contamination.
   // Two DIFFERENT runs: run X lives ONLY in the legacy/gen-0 DB, run Y lives ONLY in the
   // new/gen-1 DB. BOTH must appear in ClickHouse exactly once with their own status. This proves
   // (a) BOTH sources became leader and streamed (a single-leader regression would drop one run),

apps/webapp/test/runsReplicationService.part9.test.ts

@@ -200,7 +200,7 @@ describe("RunsReplicationService (part 9/9) - per-source replication-lag attribu
     }
   );
 
-  // Step 1.4: single-source passthrough. When a single source is used, the lag
+  // Single-source passthrough. When a single source is used, the lag
   // histogram records exactly one `source` label value (the source's id).
   replicationContainerTest(
     "records a single source label in single-source mode",