Add optimized path for power-of-two-sized buffers

[lmbarros]

Some operations on the circular buffer make use of modulo arithmetic,
which was implemented (not surprisingly!) with the modulo operator %.
It turns out that the modulo operator is quite slow and therefore, on
tight loops, a substantial portion of the run time may be spent on it.

This commit adds a separate code path for buffers with power-of-two
sizes. For these, we can implement modulo arithmetic substantially more
efficiently using some bit manipulation.

This is not a breaking change: the public interface is exactly the same
as before. An optimized buffer is transparently created under the hood
whenever the requested size is a power-of-two.

The commit also adds benchmarks that perform various operations on
buffers that are and that are not power-of-two-sized. Here's the result
of the execution on my laptop:

$go test -bench .

goos: linux
goarch: amd64
pkg: github.com/balena-os/circbuf
cpu: Intel(R) Core(TM) i7-10750H CPU @ 2.60GHz
Benchmark_Write_1024_500-12        	   73714	     15954 ns/op	31340.06 MB/s
Benchmark_Write_1025_500-12        	   54277	     22072 ns/op	22653.20 MB/s
Benchmark_Write_1024_5000-12       	   67614	     17685 ns/op	282722.48 MB/s
Benchmark_Write_1025_5000-12       	   50191	     23693 ns/op	211029.91 MB/s
Benchmark_Write_65536_5000-12      	   14248	     83924 ns/op	59577.56 MB/s
Benchmark_Write_65537_5000-12      	   13818	     86788 ns/op	57611.47 MB/s
Benchmark_Write_1024_5-12          	  170706	      6333 ns/op	 789.51 MB/s
Benchmark_Write_1025_5-12          	   80103	     14949 ns/op	 334.48 MB/s
Benchmark_WriteByte_1024-12        	  410422	      2806 ns/op	 356.44 MB/s
Benchmark_WriteByte_1025-12        	  110097	     10783 ns/op	  92.74 MB/s
Benchmark_WriteByte_65536-12       	  432852	      2831 ns/op	 353.25 MB/s
Benchmark_WriteByte_65537-12       	  108091	     10526 ns/op	  95.01 MB/s
Benchmark_Get_HalfFull_1024-12     	     920	   1291666 ns/op	   0.77 MB/s
Benchmark_Get_HalfFull_1025-12     	     908	   1310977 ns/op	   0.76 MB/s
Benchmark_Get_Full_1024-12         	     460	   2574138 ns/op	   0.39 MB/s
Benchmark_Get_Full_1025-12         	     458	   2629744 ns/op	   0.38 MB/s
Benchmark_Get_TwiceFull_1024-12    	     444	   2696897 ns/op	   0.37 MB/s
Benchmark_Get_TwiceFull_1025-12    	     132	   8926386 ns/op	   0.11 MB/s

I also tried this on some ARM devices (Pi 3B+, Pi Zero). While the exact numbers
vary, the overall picture is about the same.

[lmbarros]

FWIW, I also opened a PR to try to get all our optimizations upstream: armon#5

[robertgzr]

very nice! you will need to add a repo.yml file like this: https://github.com/balena-os/balenaos-in-container/blob/master/repo.yml
to make versionbot happy