Add configurable LOD thread limits for load, ingest, mesh, and save

[Shampun2281-tech]

Summary

• Adds separate CPU thread limits for LOD loading, chunk ingest, mesh generation, and section saving
• Settings are exposed in Sodium → Voxy and applied live without restart
• Adds Russian localization for the new options

Motivation

Previously Voxy used a single service thread pool with hardcoded service weights, which could saturate all CPU cores during LOD work. These settings let users cap parallelism per pipeline stage.

Test plan

• Built successfully with Gradle
• Verified settings appear in Sodium → Voxy
• Verified changing thread limits affects LOD workload in-game

[MCRcortex]

Dont quite understand the point of this sorry, the point of voxy's thread pool is to fully utilize all the threads in the pool.
There is a seperate issue atm in which the number of unified sodium threads that are used arnt limited to the thread config.
asside from that there are some issues with the implementation, the service weight is ment tobe immuntable, ActiveSectionTracker should not be blocked as this can very easily cause a deadlock
The service weight is hardcoded atm since its based on priority of importance to try reduce dataloss and cascading updates
if your cpu is being fully saturated, try reducing the number of threads for both sodium and voxy