Optimizations 2

[anatoly-kussul]

At this point I was just curious about how much further I can push it.
Feel free to close this PR if you feel like this is too much.

benchmarks:

goos: linux
goarch: amd64
pkg: github.com/lithammer/shortuuid/v4
cpu: 12th Gen Intel(R) Core(TM) i5-12400F
                         │     old      │                new10                │
                         │    sec/op    │   sec/op     vs base                │
UUID-12                     119.6n ± 0%   103.6n ± 1%  -13.38% (p=0.000 n=10)
Encoding-12                 42.32n ± 0%   29.57n ± 2%  -30.12% (p=0.000 n=10)
EncodingB57_MB-12           116.5n ± 2%   114.6n ± 2%        ~ (p=0.117 n=10)
EncodingB16-12              151.9n ± 0%   119.1n ± 1%  -21.59% (p=0.000 n=10)
EncodingB16_MB-12           191.2n ± 2%   189.2n ± 1%        ~ (p=0.670 n=10)
Decoding-12                167.35n ± 1%   27.08n ± 1%  -83.82% (p=0.000 n=10)
DecodingB16-12              200.0n ± 1%   230.9n ± 1%  +15.43% (p=0.000 n=10)
DecodingB16_MB-12           249.9n ± 2%   265.8n ± 1%   +6.38% (p=0.000 n=10)
NewWithAlphabet-12          381.0n ± 2%   419.0n ± 1%   +9.97% (p=0.000 n=10)
NewWithAlphabetB16-12       293.6n ± 1%   256.7n ± 2%  -12.57% (p=0.000 n=10)
NewWithAlphabetB16_MB-12    403.5n ± 1%   425.5n ± 2%   +5.45% (p=0.000 n=10)
NewWithNamespace-12         166.3n ± 1%   151.1n ± 3%   -9.17% (p=0.000 n=10)
NewWithNamespaceHttp-12     169.8n ± 1%   151.8n ± 1%  -10.66% (p=0.000 n=10)
NewWithNamespaceHttps-12    167.4n ± 1%   152.9n ± 2%   -8.60% (p=0.000 n=10)
geomean                     177.4n        146.6n       -17.38%

[lithammer]

I see that this has some overlap with #63, would you mind rebasing to make it easier to review?

[anatoly-kussul]

I see that this has some overlap with #63, would you mind rebasing to make it easier to review?

Sure, rebased

[anatoly-kussul]

Did a bit of restructuring, moving base57 encoder as separate entity,
optimized default decoding.

updated benchmarks first comment

[anatoly-kussul]

After I did restructuring, I realized that a lot of changes I made were unnecessary, so I reverted them.
Now the PR is a bit more clear.