net: precompute rfc6724policyTable in addrselect

[Thorleon]

As net package has one of the biggest init time in standard library, I have tried to improve performance by doing two things in net/addrselect.go:

1. Precompute slice with RFC rules. Currently the rules are computed and sorted in init() function. We could save the time and allocations by using prepopulated values in sorted manner. The rules haven't changed since 2015. To be extra safe we could move order validation as test case. It should slightly speed up startup of each binary with "net" package and go dns resolver. It also saves 38 allocations, ~50% of allocations in init phase of net module.
2. Replace internal net.IP usage with netip.Addr in sortByRFC6724 function. It results in ~40% performance improvement on samples from tests.

The only risk is the difference between net.IP and netip.Addr behaviour.

Init benchmark:

name    old time/op    new time/op    delta
Init-8    2.14µs ± 0%    0.12µs ± 0%   ~     (p=1.000 n=1+1)

name    old alloc/op   new alloc/op   delta
Init-8    1.05kB ± 0%    0.38kB ± 0%   ~     (p=1.000 n=1+1)

name    old allocs/op  new allocs/op  delta
Init-8      39.0 ± 0%       1.0 ± 0%   ~     (p=1.000 n=1+1)

Whole sortByRFC6724 function benchmark:

name               old time/op  new time/op  delta
SortByRFC6724/0-8   479ns ± 0%   290ns ± 0%   ~     (p=1.000 n=1+1)
SortByRFC6724/1-8   482ns ± 0%   291ns ± 0%   ~     (p=1.000 n=1+1)
SortByRFC6724/2-8   489ns ± 0%   297ns ± 0%   ~     (p=1.000 n=1+1)
SortByRFC6724/3-8   580ns ± 0%   367ns ± 0%   ~     (p=1.000 n=1+1)
SortByRFC6724/4-8   934ns ± 0%   574ns ± 0%   ~     (p=1.000 n=1+1)

[gopherbot]

Change https://go.dev/cl/419356 mentions this issue: net: precompute rfc6724policyTable in addrselect

[ianlancetaylor]

We should either have a test case verifying that the table is correct, or we should generate it with a go:generate command.

[Thorleon]

Sure @ianlancetaylor, I have added test for verifying table content correctness, in addition to test case with order checking.