Micro optimize extend

[arthurprs]

Improve code generation by only advancing the iterator in 2 locations (down from 4) and one less push/reserve location, which increases the chances of inlining and generates less code.

One benchmark improved (bench_extend_filtered), and the rest remains the same.

Before

test bench_extend                      ... bench:          37.69 ns/iter (+/- 0.62)
test bench_extend_filtered             ... bench:         143.90 ns/iter (+/- 10.50)
test bench_extend_filtered_small       ... bench:          19.27 ns/iter (+/- 3.83)
test bench_extend_from_slice           ... bench:          25.01 ns/iter (+/- 0.45)
test bench_extend_from_slice_small     ... bench:           6.43 ns/iter (+/- 0.02)
test bench_extend_from_slice_vec       ... bench:          65.69 ns/iter (+/- 2.89)
test bench_extend_from_slice_vec_small ... bench:          15.37 ns/iter (+/- 1.29)
test bench_extend_small                ... bench:          18.09 ns/iter (+/- 0.06)
test bench_extend_vec                  ... bench:          70.24 ns/iter (+/- 0.77)
test bench_extend_vec_filtered         ... bench:         128.12 ns/iter (+/- 11.17)
test bench_extend_vec_filtered_small   ... bench:          19.82 ns/iter (+/- 0.88)
test bench_extend_vec_small            ... bench:          14.65 ns/iter (+/- 1.08)

After

test bench_extend                      ... bench:          37.20 ns/iter (+/- 7.43)
test bench_extend_filtered             ... bench:         130.83 ns/iter (+/- 8.19)
test bench_extend_filtered_small       ... bench:          16.48 ns/iter (+/- 0.16)
test bench_extend_from_slice           ... bench:          24.96 ns/iter (+/- 0.49)
test bench_extend_from_slice_small     ... bench:           6.42 ns/iter (+/- 0.01)
test bench_extend_from_slice_vec       ... bench:          66.46 ns/iter (+/- 2.00)
test bench_extend_from_slice_vec_small ... bench:          15.94 ns/iter (+/- 0.43)
test bench_extend_small                ... bench:          16.63 ns/iter (+/- 0.03)
test bench_extend_vec                  ... bench:          70.70 ns/iter (+/- 1.83)
test bench_extend_vec_filtered         ... bench:         132.82 ns/iter (+/- 9.07)
test bench_extend_vec_filtered_small   ... bench:          20.70 ns/iter (+/- 0.77)
test bench_extend_vec_small            ... bench:          14.92 ns/iter (+/- 0.82)

[mbrubeck]

Thanks!