simd-0332: reduce chacha rounds for turbine from 20 to 8

[0x0ece]

• Added a few missing functions for chacha8
• Made wsample configurable with chacha8 vs chacha20 (sampler->rng_algo)
• Made sdest use chacha8 vs chacha20 based on staked wsample. Note: leader schedule (that also uses wsample) doesn’t need chacha8
• Added frank feature gate and cleaned up existing ones
• Put all together in shred tile (note: works for frank, doesn’t work yet for full client)
• Added a conformance test with Rust code to generate the C input - found a bug 🎉

Agave: anza-xyz/agave#7353

[github-actions]

Performance Measurements ⏳

Suite	Baseline	New	Change
backtest mainnet-368528500-perf per slot	0.051658 s	0.05152 s	-0.267% ✅
backtest mainnet-368528500-perf snapshot load	1.684 s	1.687 s	0.178% ✅
backtest mainnet-368528500-perf total elapsed	51.658053 s	51.519905 s	-0.267% ✅
firedancer mem usage with mainnet.toml	1005.23 GiB	1005.23 GiB	0.000% ✅

[github-actions]

Performance Measurements ⏳

Suite	Baseline	New	Change
backtest mainnet-368528500-perf per slot	0.051377 s	0.05115 s	-0.442% ✅
backtest mainnet-368528500-perf snapshot load	1.699 s	1.682 s	-1.001% ✅
backtest mainnet-368528500-perf total elapsed	51.376541 s	51.150201 s	-0.441% ✅
firedancer mem usage with mainnet.toml	1005.23 GiB	1005.23 GiB	0.000% ✅

[github-actions]

Performance Measurements ⏳

Suite	Baseline	New	Change
backtest mainnet-368528500-perf per slot	0.050528 s	0.050809 s	0.556% ✅
backtest mainnet-368528500-perf snapshot load	1.644 s	1.686 s	2.555% ✅
backtest mainnet-368528500-perf total elapsed	50.528171 s	50.808756 s	0.555% ✅
firedancer mem usage with mainnet.toml	1005.23 GiB	1005.23 GiB	0.000% ✅

[github-actions]

Performance Measurements ⏳

Suite	Baseline	New	Change
backtest mainnet-368528500-perf per slot	0.050197 s	0.05051 s	0.624% ✅
backtest mainnet-368528500-perf snapshot load	1.638 s	1.642 s	0.244% ✅
backtest mainnet-368528500-perf total elapsed	50.197126 s	50.509815 s	0.623% ✅
firedancer mem usage with mainnet.toml	993.23 GiB	993.23 GiB	0.000% ✅

[github-actions]

Performance Measurements ⏳

Suite	Baseline	New	Change
backtest mainnet-368528500-perf per slot	0.050541 s	0.050169 s	-0.736% ✅
backtest mainnet-368528500-perf snapshot load	1.626 s	1.627 s	0.062% ✅
backtest mainnet-368528500-perf total elapsed	50.541024 s	50.169485 s	-0.735% ✅
firedancer mem usage with mainnet.toml	993.23 GiB	993.23 GiB	0.000% ✅

[github-actions]

Performance Measurements ⏳

Suite	Baseline	New	Change
backtest mainnet-368528500-perf per slot	0.050317 s	0.050249 s	-0.135% ✅
backtest mainnet-368528500-perf snapshot load	1.64 s	1.632 s	-0.488% ✅
backtest mainnet-368528500-perf total elapsed	50.316589 s	50.249414 s	-0.134% ✅
firedancer mem usage with mainnet.toml	993.23 GiB	993.23 GiB	0.000% ✅