use read_u16 and fix clippy warnings

[cristicbz]

This PR adds a read_u16 step for write-s, which speeds up the small string case a little bit (see benchmarks below). It also replaces the 32-bit seed with the one from rustc, rather than using than using the last 32bits of the 64bit one.

It also fixes all clippy warnings:

• FireFox -> Firefox (correct capitalisation)
• u32::from and u64::from instead of as casts for lossless conversions.
• use separators for the seeds to make them easier to read.

Thanks for the crate and let me know if I haven't conformed to style guidelines or anything else!

Benchmarks

fx32 before:

test bench_fx32_003    ... bench:           2 ns/iter (+/- 0)
test bench_fx32_004    ... bench:           2 ns/iter (+/- 0)
test bench_fx32_011    ... bench:           5 ns/iter (+/- 0)
test bench_fx32_012    ... bench:           3 ns/iter (+/- 0)
test bench_fx32_023    ... bench:           6 ns/iter (+/- 0)
test bench_fx32_024    ... bench:           5 ns/iter (+/- 0)
test bench_fx32_068    ... bench:          14 ns/iter (+/- 0)
test bench_fx32_132    ... bench:          37 ns/iter (+/- 4)


fx32 after:

test bench_fx32_003    ... bench:           2 ns/iter (+/- 0)
test bench_fx32_004    ... bench:           2 ns/iter (+/- 0)
test bench_fx32_011    ... bench:           3 ns/iter (+/- 0)
test bench_fx32_012    ... bench:           3 ns/iter (+/- 0)
test bench_fx32_023    ... bench:           5 ns/iter (+/- 0)
test bench_fx32_024    ... bench:           5 ns/iter (+/- 0)
test bench_fx32_068    ... bench:          14 ns/iter (+/- 0)
test bench_fx32_132    ... bench:          35 ns/iter (+/- 1)


fx64 before:
test bench_fx64_003    ... bench:           3 ns/iter (+/- 0)
test bench_fx64_004    ... bench:           1 ns/iter (+/- 0)
test bench_fx64_011    ... bench:           4 ns/iter (+/- 0)
test bench_fx64_012    ... bench:           2 ns/iter (+/- 0)
test bench_fx64_023    ... bench:           5 ns/iter (+/- 0)
test bench_fx64_024    ... bench:           3 ns/iter (+/- 0)
test bench_fx64_068    ... bench:           7 ns/iter (+/- 0)
test bench_fx64_132    ... bench:          15 ns/iter (+/- 0)

fx64 after:
test bench_fx64_003    ... bench:           2 ns/iter (+/- 0)
test bench_fx64_004    ... bench:           1 ns/iter (+/- 0)
test bench_fx64_011    ... bench:           3 ns/iter (+/- 0)
test bench_fx64_012    ... bench:           3 ns/iter (+/- 0)
test bench_fx64_023    ... bench:           4 ns/iter (+/- 0)
test bench_fx64_024    ... bench:           3 ns/iter (+/- 0)
test bench_fx64_068    ... bench:           7 ns/iter (+/- 0)
test bench_fx64_132    ... bench:          15 ns/iter (+/- 2)

[Boscop]

@cbreeden Could you please publish a new release to get this improvement in the crates.io version? :)