Fix goroutine leak at silence bulk import

[moonyoungCHAE]

In silence bulk import, I noticed that if a decoding error occurs, the number of goroutines remains equal to the number of workers which is worker count. This happened because the channel wasn’t closed when an error occurred and the function exited. I fixed this issue by ensuring the channel is closed even in error cases.

I tested like this.

func TestName(t *testing.T) {
	var err error
	alertmanagerURL, err = url.Parse("http://example.com")
	if err != nil {
		log.Fatal(err)
	}

	goroutineCount := runtime.NumGoroutine()
	println("before bulk import ", goroutineCount) // print 2
	cmd := silenceImportCmd{
		force:   false,
		workers: 10,
		file:    "./test.json",
	}
	cmd.bulkImport(context.TODO(), nil)

	for i := 0; i < 100; i++ {
		goroutineCount := runtime.NumGoroutine()
		println(goroutineCount) // as-is: 13 (worker count), to-be: 2
		time.Sleep(1 * time.Second)
	}
}

[
  {
    "ID": "123",
    "Comment": "First silence",
    "CreatedBy": "user1",
    "EndsAt": 1 // this makes decode error
  },
  {
    "ID": "456",
    "Comment": "Second silence",
    "CreatedBy": "user2"
  }
]

[Spaceman1701]

LGTM, this seems reasonable. Since bulkImport returning causes the program to exit, I don't think this has much impact. However, it is a little smelly to leave these workers around and I guess some future code change could call bulkImport as part of a workflow.

[Spaceman1701]

I am 99.9% sure the test failing is unrelated. I think a real maintainer can rerun that before merging.