[dnm]: use background profiling

Background profiling is an experimental patch that was floated as a way
to get granular CPU usage information in
golang/go#23458.

I decided to dust off the Golang patch that implements this
(https://go-review.googlesource.com/c/go/+/102755) and see for myself
what you actually get. The output is something like this:

```
bgprof_test.go:51: ts: 12h37m15.540022251s
    labels: map[foo:bar]
    fn: hash/crc32.ieeeCLMUL
bgprof_test.go:51: ts: 12h37m15.54801909s
    labels: map[foo:bar]
    fn: hash/crc32.ieeeCLMUL
bgprof_test.go:51: ts: 12h37m15.560015269s
    labels: map[foo:bar]
    fn: hash/crc32.ieeeCLMUL
bgprof_test.go:51: ts: 12h37m15.568026639s
    labels: map[foo:bar]
    fn: hash/crc32.ieeeCLMUL
bgprof_test.go:51: ts: 12h37m15.580029608s
    labels: map[foo:bar]
    fn: hash/crc32.ieeeCLMUL
...
```

If we used profiler labels to identify queries (similar to what's shown
in cockroachdb#60508) I guess I can
see how you can somehow build a profile from this stream and then use
the tag breakdown of the profile to reason about CPU allocated to each
query. It seems wildly experimental and also unergonomical, though.

It seems like we'd get maybe roughly the same, without as much bending
over backwards and wildly experimental bleeding edge, by relying on
periodic foreground sampling at a lower frequency
(`runtime.SetCPUProfileRate`) than the default 100hz.

To run this PR (which you shouldn't need to do for anything, but still)
you'd clone `cockroachdb/go`, check out the [bgprof] branch, run
`./make.bash` in `src` and change your PATH so that `$(which go)` is
`bin/go`.

Release note: None

pkg/server/bgprof_test.go

@@ -0,0 +1,94 @@
+package server_test
+
+import (
+	"context"
+	"hash/crc32"
+	"runtime"
+	runtimepprof "runtime/pprof"
+	"testing"
+	"time"
+	"unsafe"
+
+	"github.com/cockroachdb/cockroach/pkg/util/leaktest"
+	stop "github.com/cockroachdb/cockroach/pkg/util/stop"
+)
+
+func TestBackgroundProfiling(t *testing.T) {
+	defer leaktest.AfterTest(t)()
+
+	ctx := context.Background()
+	//tc := testcluster.StartTestCluster(t, 3, base.TestClusterArgs{})
+	//defer tc.Stopper().Stop(ctx)
+	s := stop.NewStopper()
+	defer s.Stop(ctx)
+
+	done := make(chan struct{})
+	s.RunAsyncTask(ctx, "spin", func(ctx context.Context) {
+		runtimepprof.Do(ctx, runtimepprof.Labels("foo", "bar"), func(ctx context.Context) {
+			buf := make([]byte, 1024)
+			for {
+				select {
+				case <-done:
+					return
+				default:
+				}
+				for i := 0; i < 1000; i++ {
+					crc32.Update(0, crc32.IEEETable, buf)
+				}
+			}
+		})
+	})
+
+	runtime.EnableProfCPU(100)
+	time.Sleep(time.Second)
+	runtime.EnableProfCPU(0)
+	close(done)
+	n := 0
+	readProfile(t, runtime.ReadProfCPU, func(d time.Duration, frames *runtime.Frames, labels map[string]string) {
+		n++
+		if n < 100 {
+			fr, _ := frames.Next()
+			t.Logf("ts: %s\nlabels: %s\nfn: %s", d, labels, fr.Function)
+		}
+		return
+	})
+}
+
+func readProfile(
+	t *testing.T,
+	read func() ([]uint64, []unsafe.Pointer, bool),
+	f func(d time.Duration, frames *runtime.Frames, labels map[string]string),
+) {
+	var us []uintptr
+	for {
+		data, tags, eof := read()
+		if eof {
+			break
+		}
+		for len(data) > 0 {
+			n := int(data[0])
+			if n < 3 || n > len(data) {
+				t.Errorf("bad record: n=%d, len(data)=%d", n, len(data))
+				break
+			}
+			_ = runtime.ReadProfCPU
+			if count := data[2]; count != 0 { // 1 or num_samples or overflow (0)
+				var labelMap map[string]string
+				if tags[0] != nil {
+					labelMap = *(*map[string]string)(tags[0])
+				}
+
+				d := time.Duration(data[1]) // time since process start
+				for _, n := range data[3:n] {
+					us = append(us, uintptr(n))
+				}
+				frames := runtime.CallersFrames(us)
+				us = us[:0]
+				f(d, frames, labelMap)
+			} // else: overflow record
+
+			data = data[n:]
+			tags = tags[1:]
+		}
+	}
+}