Benchmark encoder against strings.Join style implementation

[at15]

When working on gommon errors, came back to hashicorp/go-multierrors and they use strings.Join https://github.com/hashicorp/go-multierror/blob/master/format.go#L26:20 and the implementation is a bit different style compared to how we write encoder

func Join(a []string, sep string) string {
	switch len(a) {
	case 0:
		return ""
	case 1:
		return a[0]
	case 2:
		// Special case for common small values.
		// Remove if golang.org/issue/6714 is fixed
		return a[0] + sep + a[1]
	case 3:
		// Special case for common small values.
		// Remove if golang.org/issue/6714 is fixed
		return a[0] + sep + a[1] + sep + a[2]
	}
	n := len(sep) * (len(a) - 1)
	for i := 0; i < len(a); i++ {
		n += len(a[i])
	}

	b := make([]byte, n)
	bp := copy(b, a[0])
	for _, s := range a[1:] {
		bp += copy(b[bp:], sep)
		bp += copy(b[bp:], s)
	}
	return string(b)
}

https://github.com/libtsdb/libtsdb-go/blob/master/libtsdb/common/influxdb/encoder.go

func (e *Encoder) WritePointIntTagged(p *pb.PointIntTagged) {
	e.Buf = append(e.Buf, p.Name...)
	e.Buf = append(e.Buf, ',')
	for _, tag := range p.Tags {
		e.Buf = append(e.Buf, tag.K...)
		e.Buf = append(e.Buf, '=')
		e.Buf = append(e.Buf, tag.V...)
		e.Buf = append(e.Buf, ',')
	}
	e.Buf[len(e.Buf)-1] = ' '
	e.Buf = append(e.Buf, e.DefaultField...)
	e.Buf = append(e.Buf, '=')
	e.Buf = strconv.AppendInt(e.Buf, p.Point.V, 10)
	e.Buf = append(e.Buf, ' ')
	e.Buf = strconv.AppendInt(e.Buf, p.Point.T, 10)
	e.Buf = append(e.Buf, '\n')
}

not sure which one is faster, my implementation is likely to have more allocation because the slice is large enough ... though in our case since clients are not thread safe and use just one large slice as buffer, it's pretty likely no allocation will happen ....

[at15]

It is a good use case for benchhub, so just keep it open for now ... I always spend so much time on micro optimizing ... meh.