tensorflow.load vs download URL

I am a beginner with TensorFlow 2, I am using version 2.3.1 of tensorflow.

I want to build an Image Classifier based on Inception v3. Before I can use the data in the Inception network, I have to prepare the data at first. For this task I will use the 'oxford_flower102' dataset. I found two ways to get datasets, but I don't know which way should be used in which situation.

1. by using tfds.load

import tensorflow as tf
dataset, dataset_info = tfds.load('oxford_flowers102, with_info=True, as_supervised=True)

2. by downloading the dataset with URL

from six.moves import urllib
import os
import tarfile

FILE_NAME='102flowers.tgz'
FLOWERS_DIR = os.getcwd() + '/jpg'

def download_images():
   downloadedPath = os.getcwd()+ '/'+FILE_NAME
   if not os.path.exists(downloadedPath):
       download_ulr='http://www.robots.ox.ac.uk/~vgg/data/flowers/102/102flowers.tgz'
       urllib.request.urlretrieve(download_ulr,FILE_NAME)
   else:
       print('dataset already downloaded in : ', downloadedPath)

def extract_tgz_file():
   if not os.path.exists(FLOWERS_DIR):
       print('extracting files ...')
       tar=tarfile.open(FILE_NAME, "r:gz")
       tar.extractall()
       tar.close()
       print('files now extracted in : ', FLOWERS_DIR)
   else:
       print('extraced file already exists in : ', FLOWERS_DIR)

def make_dataset_ready():
   download_images()
   extract_tgz_file()

The first way is much shorter and the split into training and test datasets is done easily. The second way is much longer and the split into training and test dataset have to be done manually. Now let me come to my questions:

1. what are the benefits of the second way, by downloading it through URL and split later the dataset manually into training and test datasets?
2. My goal will be later to feed the datasets into inception v3, are there any disadvantages for this task by using tfds.load() function ?

Thank you for the support!