modin: [PERF] Why is the first `read_csv` call slower than subsequent `read_csv` calls?

import time
import pandas
import numpy as np
import modin.pandas as pd
import modin.config as cfg
import ray

print(f"\tPandas version: {pandas.__version__}")
print(f"\tModin version: {pd.__version__}")
print(f"\tCpuCount: {cfg.CpuCount.get()}")
print(f"\tEngine: {cfg.Engine.get()}")
print(f"\tNPartitions: {cfg.NPartitions.get()}")

ray.init(runtime_env={'env_vars': {'__MODIN_AUTOIMPORT_PANDAS__': '1'}})

pandas_df = pandas.DataFrame(
    np.random.randint(0, 100, size=(1000000, 13))
)
pandas_df.to_csv("foo.csv", index=False)

def read_csv_with_pandas():
    start_time = time.time()

    pandas_df = pandas.read_csv("foo.csv", index_col=0)

    end_time = time.time()
    pandas_duration = end_time - start_time
    print("Time to read_csv with Pandas: {} seconds".format(round(pandas_duration, 3)))
    return pandas_df


def read_csv_with_modin():
    start_time = time.time()

    modin_df = pd.read_csv("foo.csv", index_col=0)

    end_time = time.time()
    modin_duration = end_time - start_time
    print("Time to read_csv with Modin: {} seconds".format(round(modin_duration, 3))) 
    return modin_df

for i in range(5):
    pandas_df = read_csv_with_pandas()
    modin_df = read_csv_with_modin()

        Pandas version: 1.5.1
        Modin version: 0.16.0+24.g11ba4811
        CpuCount: 8
        Engine: Ray
        NPartitions: 8

Time to read_csv with Pandas: 0.708 seconds
Time to read_csv with Modin: 4.132 seconds
Time to read_csv with Pandas: 0.735 seconds
Time to read_csv with Modin: 0.37 seconds
Time to read_csv with Pandas: 0.646 seconds
Time to read_csv with Modin: 0.377 seconds
Time to read_csv with Pandas: 0.673 seconds
Time to read_csv with Modin: 0.371 seconds
Time to read_csv with Pandas: 0.672 seconds
Time to read_csv with Modin: 0.379 seconds