Document generation of array types using fill; support Wrapping types

[dhardy]

I ran a few benchmarks (using thread_rng and x86_64):

# 1 kiB of u64:
test gen_1k_fill          ... bench:         335 ns/iter (+/- 24) = 3056 MB/s
test gen_1k_gen_array     ... bench:         249 ns/iter (+/- 18) = 4112 MB/s
test gen_1k_iter_repeat   ... bench:         335 ns/iter (+/- 9) = 3056 MB/s
test gen_1k_sample_iter   ... bench:         328 ns/iter (+/- 11) = 3121 MB/s
# 1 kiB of u32:
test gen_1k_fill          ... bench:         329 ns/iter (+/- 17) = 3112 MB/s
test gen_1k_gen_array     ... bench:         476 ns/iter (+/- 11) = 2151 MB/s
test gen_1k_iter_repeat   ... bench:         652 ns/iter (+/- 18) = 1570 MB/s
test gen_1k_sample_iter   ... bench:         630 ns/iter (+/- 10) = 1625 MB/s
# 1 kiB of u16:
test gen_1k_fill          ... bench:         330 ns/iter (+/- 6) = 3103 MB/s
test gen_1k_gen_array     ... bench:         933 ns/iter (+/- 135) = 1097 MB/s
test gen_1k_iter_repeat   ... bench:       1,299 ns/iter (+/- 61) = 788 MB/s
test gen_1k_sample_iter   ... bench:       1,250 ns/iter (+/- 25) = 819 MB/s

[dhardy]

CI failures because of the dyn deprecation thing and forbidding warnings in docs. Lets do another PR for that; you want to?