mono bn254 51-bit SIMD FMA multiplier

[xrvdg]

bn254 multiplier variant that uses SIMD round-to-nearest fused multiply add to compute a single multiplication. This variant is:

• the fastest single multiplier on ARM cortex, but not as high throughput as the batching variants.
• the fastest single multiplier on WASM and outperforms the batched variant batched bn254 51 bits RNE multiplier  #259 on WASM.

Merge after #259

[ocdbytes]

LGTM