[optim] Optimized RecipeCache - don't clear full cache on overflow - remove only oldest (by usage) entries

[optim] Optimized RegionFileCache - don't clear full cache on overflow - remove only oldest (by usage) entries ...

Also fixed possible deadlock

Update player chunk send queue on server coord changes (don't await client packet) ...

1. It is an optimization because we don't rebuilding chunk list on each player packet
2. It is an improvement because we don't awaiting client packet on player server coord change (like teleportation)
3. It is a fix because some client mods don't send coord update packet to server before first chu packet handling

Fixed Entity & TileEntity unloading in worlds with 0 active chunks (memory leak on world generation)

Prevent teleportation while riding entity

[optim] Optimized Chunk.scheduleBlockUpdate() ...

- use ShortObjMap with chunk coord keys instead of HashSet<PendingBlockUpdate>
  - deduplicate PendingBlockUpdate by coord
  - pool ShortObjMap and TreeSet objects

Fixed player NBT data read/write out-of-sync ...

In ultramine core player data saves asynchronously, so it was possible to read OLD data after
NEW data requested to save, but has not really been written to the file.

[compat] Fixed GalactiGreg ore generation

Fixed /custmsg possible client crash

Fixed possible crashes caused by wrong eventProxy.startEntity(entity) placement

[optim] Optimized vanilla leaves decay

[optim] Totally replaced LongHashMap & IntHashMap implementation to Koloboke collections

Fixed strip color sequences from log file

Fixed /killentity

Fixed /vanish lang

[optim] ImportChunkLoader disposing ...

Now we don't wait for finishing of writing all worlds. Waiting only for current world, then closing region files. Currently implemented for ImportChunkLoader only, should be implemented for default chunk loader in future

Better hash for in-chunk block coords

Fixed 10bec8e13a PendingBlockUpdates and EntityFallingBlock processing after chunk populating

Revert "[optim] Entity.random replaced to ThreadLocalRandom (it's really faster)" ...

This reverts commit 3aa3ad9.
(Issues with some mods that uses random seed)

I18n for duration format & restart messages

Correct error-reporting on unknown WorldType and not equal generation settings in worlds.yml and level.dat

More appropriate behavior of not splitted world mode (splitWorldDirs: false)

[refactoring] typos in class names

[optim] Entity.random replaced to ThreadLocalRandom (it's really faster)

Added PendingBlockUpdates and EntityFallingBlock processing and EntityItem removal after chunk populating ...

Newly populated chunk contains lots of PendingBlockUpdate and EntityFallingBlock, that takes lot of memory. PendingBlockUpdate processing causes more EntityFallingBlock spawning. In its turn, processing of EntityFallingBlock causes PendingBlockUpdate scheduling again. It solves by multiple iterations of processing both PendingBlockUpdate and EntityFallingBlock. Finally, we removing any EntityItem, that may be spawned at that time

Fixed /memstat "Off-Heap chunk used" more than 2gb

Fixed JLine tab completion

Issue #48 CPU utilization ...

Now we calling utilizeCPU method instead of Thread.sleep in in main server loop. In this method we can do something more useful than sleep. Currently implemented SyncServerExecutor queue processing and chunk generation

Fixed lots of issues with /genworld command

World generation limited to one chunk per tick