fleetd #811: drop ticket numbers from test comments, rule 1 covers tests too

Removed the ticket reference from a class javadoc and two section banners,
and rewrote LeadLabelRepairLoopTest's javadoc to name the behaviour the
tests protect (scheduled execution, survival of a throwing tick) instead
of the history of why they were added.

fleetd/src/test/java/dev/ltms/fleet/FleetdAssemblyLifecycleTest.java

@@ -55,7 +55,7 @@ import static org.junit.jupiter.api.Assertions.assertTrue;
  * a real {@code caffeinate} subprocess, which is not one of the resources the ticket's acceptance
  * criteria names (scheduler/inbox/mailbox/loop/MCP server/herdr router). {@code fleet.leaders:} is
  * left unset too, which is why the {@code bridge-lead-repair-} scheduler is the one "always-created"
- * scheduler this config shuts down immediately rather than leaving running (fleetd #811) — an empty
+ * scheduler this config shuts down immediately rather than leaving running — an empty
  * {@code fleet.leaders} gives the label repair nothing to do.
  */
 class FleetdAssemblyLifecycleTest {

fleetd/src/test/java/dev/ltms/fleet/lead/LeadLabelRepairLoopTest.java

@@ -11,11 +11,8 @@ import static org.junit.jupiter.api.Assertions.assertDoesNotThrow;
 import static org.junit.jupiter.api.Assertions.assertTrue;
 
 /**
- * fleetd #811 follow-up: {@code LeadLauncher.reassertLeadLabels} was unit-tested only via a direct
- * call, a scenario production never reaches — nothing scheduled it. These tests drive the real
- * {@code scheduleWithFixedDelay} path on a short, real interval rather than calling {@link
- * LeadLabelRepairLoop#tick()} directly, so a future regression that removes the scheduling call
- * again fails here too.
+ * {@link LeadLabelRepairLoop} runs its repair from the real {@code scheduleWithFixedDelay} path,
+ * on its own, and keeps running later ticks even after one tick's repair throws.
  */
 class LeadLabelRepairLoopTest {
 

fleetd/src/test/java/dev/ltms/fleet/lead/LeadLauncherTest.java

@@ -716,7 +716,7 @@ class LeadLauncherTest {
      * all) — {@link LeadLauncher#relaunch(String)} must give up after exactly {@code
      * RELAUNCH_ATTEMPTS} and must not leak any of the tabs it created along the way.
      */
-    // ── fleetd #811: the daemon re-asserts its own lead tab label ────────────────────────────────
+    // ── the daemon re-asserts its own lead tab label ──────────────────────────────────────────────
 
     /**
      * Claude Code retitling its own tab is exactly the staleness this repairs: the tab
@@ -796,7 +796,7 @@ class LeadLauncherTest {
         return herdr.calls.stream().filter(c -> c.method().equals("tab.rename")).count();
     }
 
-    // ── fleetd #811: WARN when a space holds more live leads than configured ─────────────────────
+    // ── WARN when a space holds more live leads than configured ──────────────────────────────────
 
     /** Two live agents named for the same lead, in its own space, with only one instance configured. */
     @Test