refactor: use slices.SortFunc instead of sort.Slice

[alexandear]

This PR replaces sort.Slice with slices.SortFunc in Friendly.printStatistics. See https://go-review.googlesource.com/c/go/+/626038 for reasons.

The PR adds TestFriendly_printStatistics to check that sorting works as expected, as it was not covered by any test before.

[denisvmedia]

I made an isolated benchmark, and the new method is faster:

package main

import (
	"cmp"
	"math/rand"
	"sort"
	"slices"
	"testing"
	"time"
)

type statEntry struct {
	failures int
}

func BenchmarkSortSlice(b *testing.B) {
	data := generateTestData(100000)

	b.ResetTimer()
	for i := 0; i < b.N; i++ {
		temp := make([]statEntry, len(data))
		copy(temp, data)

		sort.Slice(temp, func(i, j int) bool {
			return temp[i].failures > temp[j].failures
		})
	}
}

func BenchmarkSlicesSortFunc(b *testing.B) {
	data := generateTestData(100000)

	b.ResetTimer()
	for i := 0; i < b.N; i++ {
		temp := make([]statEntry, len(data))
		copy(temp, data)

		slices.SortFunc(temp, func(a, b statEntry) int {
			return -cmp.Compare(a.failures, b.failures)
		})
	}
}

func generateTestData(size int) []statEntry {
	rand.Seed(time.Now().UnixNano())
	data := make([]statEntry, size)
	for i := 0; i < size; i++ {
		data[i] = statEntry{failures: rand.Intn(1000000)}
	}
	return data
}

user@mac % go test -bench=. -benchtime=10s
goos: darwin
goarch: arm64
pkg: benchit
cpu: Apple M3 Pro
BenchmarkSortSlice-12         	    1440	   8164691 ns/op
BenchmarkSlicesSortFunc-12    	    1690	   7128573 ns/op
PASS
ok  	benchit	25.571s