Switch to Vec4 because of Rust compiler changes

See yoanlcq/vek#66

This ends up a bit slower than the previous baseline, but remains way
faster than falling back to the C representation.

Bench for the C repr:
AMD Ryzen 9 3900X 12-Core Processor (AMD64 Family 23 Model 113 Stepping 0)
$ cargo bench --bench benchmark -- --baseline master
tracescene/10x10x4      time:   [525.46 us 527.30 us 529.31 us]
                        change: [+74.250% +75.107% +75.981%] (p = 0.00 < 0.05)
                        Performance has regressed.