refactor(coderd): optimize chatdebug (#27129)

Adds a bund of optimizations to chatdebug:

In `coderd/x/chatd/chatdebug`:
- Adds a benchmark (excluding LLM and database stuff)
- Replaces string concatenation with strings.Builder when accumulating
stream parts (~105,000ns -> ~50,00ns)
- Removes double JSON encode in RecordingTransport (114,000ns ->
64,000ns)

In `coderd/util/strings`:
- Adds a benchmark for Truncate
- Removes unnecessary allocations in Truncate (~110,000ns -> 1,550ns in
truncation case, 1 alloc -> 0 allocs in no truncation case)

> 🤖 Claude helped with this.

coderd/util/strings/strings_test.go

@@ -1,6 +1,7 @@
 package strings_test
 
 import (
+	"bytes"
 	"fmt"
 	"testing"
 
@@ -32,6 +33,7 @@ func TestTruncate(t *testing.T) {
 		{"foo", 1, "f", nil},
 		{"foo", 0, "", nil},
 		{"foo", -1, "", nil},
+		{"", 5, "", nil},
 		{"foo bar", 7, "foo bar", []strings.TruncateOption{strings.TruncateWithEllipsis}},
 		{"foo bar", 6, "foo b…", []strings.TruncateOption{strings.TruncateWithEllipsis}},
 		{"foo bar", 5, "foo …", []strings.TruncateOption{strings.TruncateWithEllipsis}},
@@ -81,6 +83,28 @@ func TestTruncate(t *testing.T) {
 	}
 }
 
+func BenchmarkTruncate(b *testing.B) {
+	b.Run("NoTruncationNeeded", func(b *testing.B) {
+		s := "a short string well under the limit"
+		b.ReportAllocs()
+		for b.Loop() {
+			strings.Truncate(s, 1000)
+		}
+	})
+
+	b.Run("ActualTruncation", func(b *testing.B) {
+		var buf bytes.Buffer
+		for range 2000 {
+			buf.WriteString("日本語テスト word ")
+		}
+		s := buf.String()
+		b.ReportAllocs()
+		for b.Loop() {
+			strings.Truncate(s, 100, strings.TruncateWithEllipsis, strings.TruncateWithFullWords)
+		}
+	})
+}
+
 func TestUISanitize(t *testing.T) {
 	t.Parallel()