Timeseries Collections in Mongodb - granularity and metaField

Hi All,
I am not very much sure on how granularity is been used in time series collections in mongodb.
If you look about below example, test_ts collection is created with granularity of seconds and 4 different objects are inserted to the timeseries collection, each with different timestamp (each measurement is taken per seconds).

db.createCollection(
    "test_ts",
    {
       timeseries: {
          timeField: "timestamp",
          metaField: "metadata",
          granularity: "seconds"
       }
    },
)
db.test_ts.insertMany( [
   {
      "metadata": { "sensorId": 5578, "type": "temperature" },
      "timestamp": ISODate("2021-05-18T00:00:00.000Z"),
      "temp": 12
   },
   {
      "metadata": { "sensorId": 5578, "type": "temperature" },
      "timestamp": ISODate("2021-05-18T00:00:01.000Z"),
      "temp": 11
   },
   {
      "metadata": { "sensorId": 5578, "type": "temperature" },
      "timestamp": ISODate("2021-05-18T00:00:02.000Z"),
      "temp": 11
   },
   {
      "metadata": { "sensorId": 5578, "type": "temperature" },
      "timestamp": ISODate("2021-05-18T00:00:03.000Z"),
      "temp": 12
   },
   {
      "metadata": { "sensorId": 5578, "type": "temperature" },
      "timestamp": ISODate("2021-05-18T00:00:04.000Z"),
      "temp": 16
   }
] )

I expected that, 4 different documents will be created in db.system.buckets.test_ts collection, as each object inserted varies with time of 1 second and granularity is seconds. But only 1 document is created.

> db.system.buckets.test_ts.countDocuments({})
1

printing the document shows below

> db.system.buckets.test_ts.find().pretty()
{
	"_id" : ObjectId("60a303809008706b3bfa88ae"),
	"control" : {
		"version" : 1,
		"min" : {
			"_id" : ObjectId("6440cfed2b3847307ef4d91a"),
			"timestamp" : ISODate("2021-05-18T00:00:00Z"),
			"temp" : 11
		},
		"max" : {
			"_id" : ObjectId("6440cfed2b3847307ef4d91e"),
			"timestamp" : ISODate("2021-05-18T00:00:04Z"),
			"temp" : 16
		}
	},
	"meta" : {
		"sensorId" : 5578,
		"type" : "temperature"
	},
	"data" : {
		"_id" : {
			"0" : ObjectId("6440cfed2b3847307ef4d91a"),
			"1" : ObjectId("6440cfed2b3847307ef4d91b"),
			"2" : ObjectId("6440cfed2b3847307ef4d91c"),
			"3" : ObjectId("6440cfed2b3847307ef4d91d"),
			"4" : ObjectId("6440cfed2b3847307ef4d91e")
		},
		"timestamp" : {
			"0" : ISODate("2021-05-18T00:00:00Z"),
			"1" : ISODate("2021-05-18T00:00:01Z"),
			"2" : ISODate("2021-05-18T00:00:02Z"),
			"3" : ISODate("2021-05-18T00:00:03Z"),
			"4" : ISODate("2021-05-18T00:00:04Z")
		},
		"temp" : {
			"0" : 12,
			"1" : 11,
			"2" : 11,
			"3" : 12,
			"4" : 16
		}
	}
}

We can see, metaField is been used to decide on number of documents here, all objects with same metaField clubbed together in one document. If granularity is not used to decide on how many objects should be merged together in one document then whats the exact use of granularity?

Also, how one should choose the granularity for highly frequent data (coming in every seconds)

Thank you in advance.