Improve mesh relaying, presence, and DM robustness; signed public msgs; reachability UI; quicker announces; store-and-forward (#527)

* Sign public broadcasts; verify relayed messages via persisted signing keys; keep scheduled relays in sparse graphs and speed their jitter; persist announce signing key for offline auth; add short backoff after disconnect errors to reduce reconnect thrash

* Add connected vs reachable model: retain peers after link drop, expire after reachability window; expose all peers in snapshots; compute isReachable in UI; add meshReachable state and sorting; avoid removing peers on link events; notify UI on stale removals

* ContentView: handle new .meshReachable connection state in header icon switch (exhaustive switch fix)

* Logs: tag relayed announces as 'Reachable via mesh' and annotate public message logs with (direct|mesh) path for easier field analysis

* Fix syntax error: remove stray else/log inserted into writeOrEnqueue; keep logs clean

* UI: use 'point.3.connected.trianglepath.dotted' for mesh-reachable; change people count to include connected+reachable (exclude Nostr-only)

* UI: switch to 'point.3.filled.connected.trianglepath.dotted' for mesh-reachable icons in list and header

* Reachability: reduce retention to 21s for all peers (verified and unverified) to minimize stale presence

* mesh DMs/acks: route to reachable peers; queue READ/DELIVERED until handshake; add Transport.isPeerReachable; UI: hide offline non-mutuals; DM header: better name fallback + show transport + encryption icons; fix NostrTransport conformance

* Verification sheet: compute encryption status and fingerprint using short mesh ID mapping (fix 'not encrypted/handshake' for DMs with stable key)

* Announce cadence: faster discovery (4s), sparse 15±4s, dense 30±8s; initial 0.6s; post-subscribe 50ms; min-force 150ms; maintenance 5s; proactive announces on handshake + recent-traffic nudge

* Relay: increase broadcast TTL cap in sparse graphs to 6; tighten jitter for handshake (10–35ms) and directed (20–60ms) relays

* Range/robustness: store-and-forward for directed packets (15s) with flush on new links + periodic; announces: no subset + afterglow re-announce on first-seen; adaptive scanning: force ON when <=2 neighbors or recent traffic

* Fix warnings: remove unused msgID and unused mutable var in directed spool flush

* Announces: TTL 7 (sparse only) via RelayController; no fanout subset for announces; neighbor-change rebroadcast of last 2–3 announces. Fragments: faster pacing (5ms global, 4ms directed).

* Peer list: real-time icon updates by publishing snapshots on connectivity checks; add unread message indicator (envelope) next to peers with unread DMs

* UI: unread envelope uses orange; hasUnreadMessages checks Nostr conv key for peers with known Nostr pubkeys (geohash DM consistency)

* Logs/robustness: debounce disconnect notifications (1.5s), debounce 'reconnected' logs (2s), add weak-link cooldown after timeouts on very weak RSSI (<= -90)

* Peer icons: faster, accurate reachability\n\n- Run connectivity checks every maintenance tick (5s)\n- Publish peer snapshots on central unsubscribe for instant UI refresh\n- Lower inactivity timeout to 8s and disconnect debounce to 0.9s\n- Gate reachability on mesh-attached (>=1 direct link); no links => no reachable peers\n- Keep 21s retention for verified/unverified, but only when attached to mesh\n\nImproves list responsiveness when walking out of range and prevents stale 'reachable' states when isolated.

---------

Co-authored-by: jack <jackjackbits@users.noreply.github.com>

bitchat/Services/RelayController.swift

@@ -15,6 +15,7 @@ struct RelayController {
                        isDirectedEncrypted: Bool,
                        isDirectedFragment: Bool,
                        isHandshake: Bool,
+                       isAnnounce: Bool,
                        degree: Int,
                        highDegreeThreshold: Int) -> RelayDecision {
         // Suppress obvious non-relays
@@ -25,7 +26,8 @@ struct RelayController {
             // Always relay with no TTL cap for these types
             let newTTL = (ttl &- 1)
             // Slight jitter to desynchronize without adding too much latency
-            let delayRange: ClosedRange<Int> = isHandshake ? 20...60 : 40...120
+            // Tighter for faster multi-hop handshakes and directed DMs
+            let delayRange: ClosedRange<Int> = isHandshake ? 10...35 : 20...60
             let delayMs = Int.random(in: delayRange)
             return RelayDecision(shouldRelay: true, newTTL: newTTL, delayMs: delayMs)
         }
@@ -42,15 +44,22 @@ struct RelayController {
         let prob = baseProb
         let shouldRelay = Double.random(in: 0...1) <= prob
 
-        // TTL clamping in dense graphs (only for broadcast)
-        let ttlCap: UInt8 = degree >= highDegreeThreshold ? 3 : 5
+        // TTL clamping for broadcast
+        // - Dense graphs: keep very low to avoid floods
+        // - Sparse graphs: allow slightly longer reach for multi-hop discovery
+        // - Announces in sparse graphs get a bit more headroom
+        let ttlCap: UInt8 = {
+            if degree >= highDegreeThreshold { return 3 }
+            return isAnnounce ? 7 : 6
+        }()
         let clamped = max(1, min(ttl, ttlCap))
         let newTTL = clamped &- 1
 
         // Wider jitter window to allow duplicate suppression to win more often
+        // For sparse graphs (<=2), relay quickly to avoid cancellation races
         let delayMs: Int
         switch degree {
-        case 0...2: delayMs = Int.random(in: 40...100)
+        case 0...2: delayMs = Int.random(in: 10...40)
         case 3...5: delayMs = Int.random(in: 60...150)
         case 6...9: delayMs = Int.random(in: 80...180)
         default:    delayMs = Int.random(in: 100...220)