Merge pull request #30 from torlando-tech/fix/convo-timestamp-clock-race-and-load-perf

fix: correct conversation timestamps (clock race) + ~2x faster message load

platformio.ini

@@ -72,10 +72,17 @@ lib_deps =
     ; and the bz2 layer + decompress-on-receive in Resource::assemble(). The
     ; far-diverged 0.3.0 fork's Resource/Transport/Identity work is subsumed by
     ; upstream's reimplementation. Bump in lockstep with microLXMF below.
-    https://github.com/torlando-tech/microReticulum.git#2f21feef9ff30ebde86370b56335813232a0d3e1
+    ; cb71ace: OS::ltime() now uses the monotonic 64-bit esp_timer instead of a
+    ; racy 32-bit millis() rollover counter (static low32/high32 raced across
+    ; FreeRTOS tasks, inflating OS::time() by N*49.7 days — garbage convo
+    ; timestamps, e.g. "Future"/"49w ago"/year-2046 now).
+    https://github.com/torlando-tech/microReticulum.git#cb71aceb8a9c58444521d5ce24ea2db10191d92d
     ; microLXMF: chore/microreticulum-0.4.1-layout — includes namespaced to
     ; <microReticulum/...> for the 0.4.x src/microReticulum/ layout.
-    https://github.com/torlando-tech/microLXMF.git#33760d0d38d81a9fc3bc07693ffeb7dd678a414c
+    ; 3cdde79: faster load_message_metadata — single LittleFS open (read_file
+    ; returns 0 on miss, no separate file_exists probe) + JSON field filter that
+    ; skips the large "packed" blob. ~2x quicker conversation load.
+    https://github.com/torlando-tech/microLXMF.git#3cdde79172af915c4e330be36ff9775550e57fa5
     lvgl/lvgl@^8.3.11
     bblanchon/ArduinoJson@^7.4.2
     hideakitai/MsgPack@^0.4.2