normalize initd state modeling and stabilize synthetic initd PID mapping

add IPC backpressure/drop telemetry and jittered swarm retry tiers
harden socket bridge timeout/lifecycle handling and replace recv polling with readiness signaling
improve shell background child isolation and worker offload fallback/output diagnostics
add var-log fast append path to reduce read-concat-write amplification

packages/bare-os-booter/lib/bare-os-swarm-peer-policy.js

@@ -62,7 +62,9 @@ function readSwarmPolicyEnv(env) {
  *   banUntilMs: number,
  *   ewmaLatencyMs: number,
  *   lastLatencyMs: number | null,
- *   reconnectBudget: number
+ *   reconnectBudget: number,
+ *   earliestRetryAtMs: number,
+ *   retryTier: 'none' | 'short' | 'medium' | 'long' | 'xlong'
  * }} PeerState
  */
 
@@ -99,13 +101,33 @@ export class BareOsSwarmPeerPolicyEngine {
         banUntilMs: 0,
         ewmaLatencyMs: 0,
         lastLatencyMs: null,
-        reconnectBudget: 32
+        reconnectBudget: 32,
+        earliestRetryAtMs: 0,
+        retryTier: 'none'
       }
       this._peers.set(k, st)
     }
     return st
   }
 
+  _retryDelayFor(st) {
+    const n = Math.max(0, Number(st.consecutiveFail) || 0)
+    const tier =
+      n >= 9 ? 'xlong' : n >= 6 ? 'long' : n >= 4 ? 'medium' : n >= 2 ? 'short' : 'none'
+    const base =
+      tier === 'xlong'
+        ? 15000
+        : tier === 'long'
+          ? 8000
+          : tier === 'medium'
+            ? 3000
+            : tier === 'short'
+              ? 800
+              : 0
+    const jitter = base > 0 ? Math.floor(base * 0.3 * Math.random()) : 0
+    return { tier, delayMs: base + jitter }
+  }
+
   /**
    * @param {string} peerKey
    * @param {boolean} ok
@@ -119,6 +141,8 @@ export class BareOsSwarmPeerPolicyEngine {
     if (ok) {
       st.ok++
       st.consecutiveFail = 0
+      st.retryTier = 'none'
+      st.earliestRetryAtMs = 0
       const lat =
         typeof probeMeta.latencyMs === 'number' && Number.isFinite(probeMeta.latencyMs)
           ? Math.max(0, probeMeta.latencyMs)
@@ -137,6 +161,9 @@ export class BareOsSwarmPeerPolicyEngine {
     } else {
       st.fail++
       st.consecutiveFail++
+      const retry = this._retryDelayFor(st)
+      st.retryTier = retry.tier
+      st.earliestRetryAtMs = now + retry.delayMs
       if (st.consecutiveFail >= this._cfg.failsBeforeBan) {
         st.banCount++
         const base = Math.min(
@@ -146,6 +173,8 @@ export class BareOsSwarmPeerPolicyEngine {
         const jitter = Math.floor(base * 0.2 * Math.random())
         st.banUntilMs = now + base + jitter
         st.consecutiveFail = 0
+        st.retryTier = 'xlong'
+        st.earliestRetryAtMs = st.banUntilMs
       }
     }
   }
@@ -174,6 +203,7 @@ export class BareOsSwarmPeerPolicyEngine {
     const now = Date.now()
     if (!st) return true
     if (st.banUntilMs > now) return false
+    if (st.earliestRetryAtMs > now) return false
     if (st.reconnectBudget <= 0) return false
     return true
   }
@@ -251,7 +281,9 @@ export class BareOsSwarmPeerPolicyEngine {
         id,
         ...st,
         banned: st.banUntilMs > now,
-        banRemainingMs: st.banUntilMs > now ? st.banUntilMs - now : 0
+        banRemainingMs: st.banUntilMs > now ? st.banUntilMs - now : 0,
+        retryWaitMs:
+          st.earliestRetryAtMs > now ? st.earliestRetryAtMs - now : 0
       })),
       atMs: now
     }