fix: ensure chunk is loaded before placing death chest block

[wangzhizhou]

Problem

When a player disconnects immediately after dying (e.g. server lag, accidental quit, or death while logging out), the death chest is never created and all items are lost forever.

Root cause

BlockCreationChestListener.onCreate() schedules the actual setType(CHEST) call on the next tick via runTask() (intentionally, to avoid the chest being destroyed by a delayed explosion when sleeping in the nether).

Meanwhile, SpawnChestListener.onDeath() clears event.getDrops() synchronously in the same death event.

If the player disconnects between these two moments:

1. Drops are already cleared (items gone from the world)
2. On the next tick, the chunk containing the death location is unloaded (no players nearby), so location.getBlock().setType(CHEST) writes to an unloaded chunk and is silently discarded
3. Result: no chest, no drops → items permanently lost

Fix

In BlockCreationChestListener.onCreate(), before placing the chest block:

• Check if the chunk is loaded; if not, force-load it via world.getChunkAt(chunkX, chunkZ)
• If it still isn't loaded (async load), retry every tick for up to 40 ticks (2 seconds)
• Only then place the block

This preserves the original next-tick delay (nether explosion protection) while guaranteeing the block lands on a loaded chunk.

Verification

On a Paper 26.2 test server, the repro is: give items → /minecraft:kill → disconnect within 2s → check chest at death location.

Scenario	Before fix	After fix
Death + immediate disconnect	❌ No chest, items lost	✅ Chest created, items intact
Death while staying online	✅ Chest created	✅ Chest created (no regression)

8/8 repeated disconnect-on-death runs passed with the fix (100% chest creation, 0 item loss), vs 100% failure before.

Also added a debug log line with the chest location to confirm placement.

[wangzhizhou]

Found a Bug And Fixed it, please review this PR, maybe this plugin can be publish on hangar or modrinth platform, and make more people can see it.

[wangzhizhou]

Additional fix (commit 3572396): destroyChest idempotency

Second issue found during full regression: on server startup, expired chests get loaded and their expiration tasks run racing with the model being added to the loaded cache. loadedChests.remove() then returns null and destroyChest throws IllegalArgumentException: Invalid model (one WARN per restart).

Fix: return silently when the model is no longer in the cache — destroying a chest is idempotent. Verified: 2 consecutive restarts with 11 expired chests → 0 exceptions (was 1/1 before).

[wangzhizhou]

Superseded by #102 (#102) which contains the complete fix set (chunk load before chest placement + idempotent destroyChest) from the fork's main branch. Closing this duplicate to keep the review focused on a single PR.