caddyhttp: presize buffers in hot paths to reduce allocations

Right-size three per-request allocations that previously grew from a
nil/empty backing array, causing repeated reallocation:

- CleanPath: pre-Grow the strings.Builder used when slash collapsing is
  disabled, instead of letting it reallocate as runes are appended
- headers: presize the rewritten value slice in HeaderOps.ApplyTo's Set
  loop instead of appending from nil
- reverseproxy: presize the weights slice in WeightedRoundRobinSelection
  .Select

This commit also adds benchmarks exercising each path. Measured with benchstat
with count=20 and GOMAXPROCS=1, with p=0.000:

                                    sec/op                  B/op              allocs/op
CleanPathNoCollapse       1013.4n -> 828.6n  (-18.23%)  120->64  (-46.67%)   4->1  (-75.00%)
HeaderOpsApplyToSet       1262.0n -> 955.8n  (-24.27%)  255->152 (-40.39%)   9->6  (-33.33%)
WeightedRoundRobinSelect   215.8n -> 200.3n  ( -7.20%)  112->96  (-14.29%)   2->2  (N/A)

modules/caddyhttp/caddyhttp.go

@@ -286,6 +286,7 @@ func CleanPath(p string, collapseSlashes bool) string {
 	// and then remove the remaining temporary characters.
 	const tmpCh = 0xff
 	var sb strings.Builder
+	sb.Grow(len(p) + 1)
 	for i, ch := range p {
 		if ch == '/' && i > 0 && p[i-1] == '/' {
 			sb.WriteByte(tmpCh)

modules/caddyhttp/caddyhttp_bench_test.go

@@ -0,0 +1,11 @@
+package caddyhttp
+
+import "testing"
+
+func BenchmarkCleanPathNoCollapse(b *testing.B) {
+	const p = "/foo/bar/baz/qux/some/longer/path/segment/here/index.html"
+	b.ReportAllocs()
+	for b.Loop() {
+		_ = CleanPath(p, false)
+	}
+}

modules/caddyhttp/headers/headers.go

@@ -243,11 +243,11 @@ func (ops *HeaderOps) ApplyTo(hdr http.Header, repl *caddy.Replacer) {
 	// set
 	for fieldName, vals := range ops.Set {
 		fieldName = repl.ReplaceKnown(fieldName, "")
-		var newVals []string
+		// use a new slice so we don't overwrite
+		// the original values in ops.Set
+		newVals := make([]string, len(vals))
 		for i := range vals {
-			// append to new slice so we don't overwrite
-			// the original values in ops.Set
-			newVals = append(newVals, repl.ReplaceKnown(vals[i], ""))
+			newVals[i] = repl.ReplaceKnown(vals[i], "")
 		}
 		hdr.Set(fieldName, strings.Join(newVals, ","))
 	}

modules/caddyhttp/headers/headers_bench_test.go

@@ -0,0 +1,24 @@
+package headers
+
+import (
+	"net/http"
+	"testing"
+
+	"github.com/caddyserver/caddy/v2"
+)
+
+func BenchmarkHeaderOpsApplyToSet(b *testing.B) {
+	ops := &HeaderOps{
+		Set: http.Header{
+			"Content-Type":    []string{"text/html; charset=utf-8"},
+			"Cache-Control":   []string{"public", "max-age=3600", "immutable"},
+			"X-Custom-Header": []string{"value-one", "value-two"},
+		},
+	}
+	repl := caddy.NewReplacer()
+	b.ReportAllocs()
+	for b.Loop() {
+		hdr := make(http.Header)
+		ops.ApplyTo(hdr, repl)
+	}
+}

modules/caddyhttp/reverseproxy/selectionpolicies.go

@@ -149,7 +149,7 @@ func (r *WeightedRoundRobinSelection) Select(pool UpstreamPool, _ *http.Request,
 		return pool[0]
 	}
 	var index, totalWeight int
-	var weights []int
+	weights := make([]int, 0, len(r.Weights))
 
 	for _, w := range r.Weights {
 		if w > 0 {

modules/caddyhttp/reverseproxy/selectionpolicies_bench_test.go

@@ -0,0 +1,25 @@
+package reverseproxy
+
+import (
+	"net/http"
+	"testing"
+)
+
+func BenchmarkWeightedRoundRobinSelect(b *testing.B) {
+	pool := UpstreamPool{
+		{Host: new(Host), Dial: "0.0.0.1"},
+		{Host: new(Host), Dial: "0.0.0.2"},
+		{Host: new(Host), Dial: "0.0.0.3"},
+		{Host: new(Host), Dial: "0.0.0.4"},
+		{Host: new(Host), Dial: "0.0.0.5"},
+	}
+	wrrPolicy := &WeightedRoundRobinSelection{
+		Weights:     []int{5, 4, 3, 2, 1},
+		totalWeight: 15,
+	}
+	req, _ := http.NewRequest("GET", "/", nil)
+	b.ReportAllocs()
+	for b.Loop() {
+		_ = wrrPolicy.Select(pool, req, nil)
+	}
+}