Improve Random Block Selection RNG

[ScriptLineStudios]

The RNG used in selecting random blocks for updates, is fundamentally flawed, plotting some values sampled from this RNG clearly indicates the problem:

Modifying the algorithm to include a couple more terms yields a much nicer and random looking result:

Additionally this PR removes the deprecated threadSafeRandom object used for providing the client with seeds for sound variant selection in favor of using the improved randValue :D

[lynxplay]

Thank you for the PR.

We are not going to deviate from vanilla's logic at the whim that it "looks more random".
Mojangs usage of the deprecated annotation is up to them, as long as they use it, there is 0 reason for us to "replace it"

[vadcx]

I remember some fork a while ago attempted to optimize block update RNG at some point.

@ScriptLineStudios If you can prove that this RNG pattern causes problems, you would need to raise issues with Mojang directly? Otherwise it must have been a deliberate choice, the same magic number is used here as an increment:

• https://en.wikipedia.org/wiki/Linear_congruential_generator#Parameters_in_common_use 1013904223
• and here's fun a reading on LFSR https://jacopretorius.net/2017/09/wolfeinstein-3d-fizzlefade-algorithm.html

[Spottedleaf]

Hi,

I know that this PR is old but I wanted to clarify something.

Paper uses an LCG random for random block selection during random ticking. Specifically, we take the lower 12 bits of Random#nextInt().

See https://github.com/Tuinity/Moonrise/blob/mc/1.21.3/src/main/java/ca/spottedleaf/moonrise/mixin/random_ticking/ServerLevelMixin.java

If you find an issue with this implementation, you may open an issue on https://github.com/Tuinity/Moonrise.