Add non-copying version of murmur3_128 that reads directly from a byte buffer

[ShiKaiWi]

Close #20

Refering to the implementation of murmur3_32_of_slice, provide murmur3_x64_128_of_slice and murmur3_x86_128_of_slice.

And the benchmark shows the improvement:

test bench_32            ... bench:          69.78 ns/iter (+/- 0.65) = 811 MB/s
test bench_32_slice      ... bench:           5.28 ns/iter (+/- 0.05) = 11200 MB/s
test bench_c_32          ... bench:          17.34 ns/iter (+/- 0.15) = 3294 MB/s
test bench_c_x64_128     ... bench:          10.94 ns/iter (+/- 0.11) = 5600 MB/s
test bench_c_x86_128     ... bench:          19.61 ns/iter (+/- 0.33) = 2947 MB/s
test bench_x64_128       ... bench:          24.23 ns/iter (+/- 0.35) = 2333 MB/s
test bench_x64_128_slice ... bench:           9.78 ns/iter (+/- 0.12) = 6222 MB/s
test bench_x86_128       ... bench:          27.90 ns/iter (+/- 0.57) = 2074 MB/s
test bench_x86_128_slice ... bench:          16.11 ns/iter (+/- 0.73) = 3500 MB/s

[ShiKaiWi]

@stusmall Hi, would you please take a look at this changeset?

[stusmall]

This is returning a Result, but I believe the operation is infallable. Is that correct? Converting this to result u128 would probably be my only change

[ShiKaiWi]

@stusmall Thanks for your review, which makes sense. I've addressed it. PTAL again. ❤️

[ShiKaiWi]

@stusmall Hi, the requested changes have been finished. Would you please take a look again?

[stusmall]

Sorry about the slow reply. Work has been chaotic and I wanted to set aside to give a more careful read. This looks great

I'm going to do some changes in the coming days to upgrade deps and add no_std support. I'll have that go out in that version