Fix global revive scheduler execution and reload handling

[Nergly]

Motivation

• The global revive task previously ran its reset logic from an async callback which can access Bukkit state off the main thread and silently fail, preventing players from being unbanned/revived at the scheduled time.
• Reloading the plugin via initialize() did not recreate the GlobalReviveUnDeathBan runnable, so schedule/config changes from /ah reload were not applied.
• Centralize lifecycle handling so the global revive scheduler is started/stopped deterministically and honors config changes.

Description

• Changed GlobalReviveUnDeathBan.run() to compute the next run, load ServerData asynchronously, and then dispatch runGlobalReset back onto the Bukkit main thread with Bukkit.getScheduler().runTask(...).
• Added explicit error handling/logging for async server-data load failures and exceptions thrown while executing reset logic on the main thread.
• Moved creation/start of GlobalReviveUnDeathBan into initialize() and removed the duplicate startup from onEnable(), and now initialize() stops any existing instance before recreating it so /ah reload updates the schedule.
• Adjusted logging so completion is logged after the actual reset work runs on the main thread.

Testing

• Built the project successfully: mvn -q -DskipTests package (build completed without errors).

---

Codex Task