Check if this solution meets your requirements.

Explanation

1. We join over the same collection. So for each item _i we take item _i+1. This method gives us where presenceStatus has been changed.
2. We filter document_{i i+1} pairs where presenceStatus is 0 - 1 or 1 - 0.
3. We group them into single data array.
4. Now we iterate over data by 2 steps (i=0;i<data.length;i+=2) and take updatedAt value.

    var occupiedTime = data[i].tmp.updatedAt
    var vacantTime   = data[i+1].tmp.updatedAt

5. We flatten calculated values and restore original document structure.

db.collection.aggregate([
  {
    $lookup: {
      from: "collection",
      let: {
        root_id: "$_id"
      },
      pipeline: [
        {
          $match: {
            $expr: {
              $gt: [
                "$_id",
                "$$root_id"
              ]
            }
          }
        },
        {
          $limit: 1
        }
      ],
      as: "tmp"
    }
  },
  {
    $match: {
      $or: [
        {
          "presenceStatus": 1,
          "tmp.presenceStatus": 0
        },
        {
          "presenceStatus": 0,
          "tmp.presenceStatus": 1
        }
      ]
    }
  },
  {
    $group: {
      _id: null,
      data: {
        $push: {
          $mergeObjects: [
            "$$ROOT",
            {
              tmp: {
                $arrayElemAt: [
                  "$tmp",
                  0
                ]
              }
            }
          ]
        }
      }
    }
  },
  {
    $addFields: {
      data: {
        $map: {
          input: {
            $range: [
              0,
              {
                $size: "$data"
              },
              2
            ]
          },
          as: "idx",
          in: {
            "occupiedTime": {
              $arrayElemAt: [
                "$data.tmp.updatedAt",
                {
                  $cond: [
                    {
                      $eq: [
                        {
                          $arrayElemAt: [
                            "$data.tmp.presenceStatus",
                            "$$idx"
                          ]
                        },
                        1
                      ]
                    },
                    "$$idx",
                    {
                      $add: [
                        "$$idx",
                        1
                      ]
                    }
                  ]
                }
              ]
            },
            "vacantTime": {
              $arrayElemAt: [
                "$data.tmp.updatedAt",
                {
                  $cond: [
                    {
                      $eq: [
                        {
                          $arrayElemAt: [
                            "$data.tmp.presenceStatus",
                            "$$idx"
                          ]
                        },
                        0
                      ]
                    },
                    "$$idx",
                    {
                      $add: [
                        "$$idx",
                        1
                      ]
                    }
                  ]
                }
              ]
            },
            "created": {
              $arrayElemAt: [
                "$data.tmp.createdAt",
                "$$idx"
              ]
            },
            "_id": {
              $arrayElemAt: [
                "$data.tmp._id",
                "$$idx"
              ]
            },
            "__v": 0
          }
        }
      }
    }
  },
  {
    $unwind: "$data"
  },
  {
    $replaceRoot: {
      newRoot: "$data"
    }
  },
  {
    $addFields: {
      "dwellTime": {
        $dateToString: {
          date: {
            $toDate: {
              $subtract: [
                "$vacantTime",
                "$occupiedTime"
              ]
            }
          },
          format: "%H-%M-%S"
        }
      }
    }
  }
])

MongoPlayground