[ci] load_boston has been removed in sklearn 1.2

[shiyu1994]

Description

The Boston dataset has been removed in sklearn 1.2 due to ethical issues. However, our test cases use the dataset for many times. The removal causes our CI jobs to fail.

Reproducible example

(https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879)

>           raise ImportError(msg)
[3900](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3901)
E           ImportError: 
[3901](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3902)
E           `load_boston` has been removed from scikit-learn since version 1.2.
[3902](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3903)
E           
[3903](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3904)
E           The Boston housing prices dataset has an ethical problem: as
[3904](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3905)
E           investigated in [1], the authors of this dataset engineered a
[3905](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3906)
E           non-invertible variable "B" assuming that racial self-segregation had a
[3906](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3907)
E           positive impact on house prices [2]. Furthermore the goal of the
[3907](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3908)
E           research that led to the creation of this dataset was to study the
[3908](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3909)
E           impact of air quality but it did not give adequate demonstration of the
[3909](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3910)
E           validity of this assumption.
[3910](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3911)
E           
[3911](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3912)
E           The scikit-learn maintainers therefore strongly discourage the use of
[3912](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3913)
E           this dataset unless the purpose of the code is to study and educate
[3913](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3914)
E           about ethical issues in data science and machine learning.
[3914](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3915)
E           
[3915](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3916)
E           In this special case, you can fetch the dataset from the original
[3916](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3917)
E           source::
[3917](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3918)
E           
[3918](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3919)
E               import pandas as pd
[3919](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3920)
E               import numpy as np
[3920](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3921)
E           
[3921](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3922)
E               data_url = "http://lib.stat.cmu.edu/datasets/boston"
[3922](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3923)
E               raw_df = pd.read_csv(data_url, sep="\s+", skiprows=22, header=None)
[3923](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3924)
E               data = np.hstack([raw_df.values[::2, :], raw_df.values[1::2, :2]])
[3924](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3925)
E               target = raw_df.values[1::2, 2]
[3925](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3926)
E           
[3926](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3927)
E           Alternative datasets include the California housing dataset and the
[3927](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3928)
E           Ames housing dataset. You can load the datasets as follows::
[3928](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3929)
E           
[3929](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3930)
E               from sklearn.datasets import fetch_california_housing
[3930](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3931)
E               housing = fetch_california_housing()
[3931](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3932)
E           
[3932](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3933)
E           for the California housing dataset and::
[3933](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3934)
E           
[3934](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3935)
E               from sklearn.datasets import fetch_openml
[3935](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3936)
E               housing = fetch_openml(name="house_prices", as_frame=True)
[3936](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3937)
E           
[3937](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3938)
E           for the Ames housing dataset.
[3938](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3939)
E           
[3939](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3940)
E           [1] M Carlisle.
[3940](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3941)
E           "Racist data destruction?"
[3941](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3942)
E           <https://medium.com/@docintangible/racist-data-destruction-113e3eff54a8>
[3942](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3943)
E           
[3943](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3944)
E           [2] Harrison Jr, David, and Daniel L. Rubinfeld.
[3944](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3945)
E           "Hedonic housing prices and the demand for clean air."
[3945](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3946)
E           Journal of environmental economics and management 5.1 (1978): 81-102.
[3946](https://github.com/microsoft/LightGBM/actions/runs/3654388214/jobs/6174767879#step:5:3947)
E           <https://www.researchgate.net/publication/4974606_Hedonic_housing_prices_and_the_demand_for_clean_air>

[jameslamb]

Thanks @shiyu1994 , but we already have an issue for this (#4793) and a PR to fix it (#5581).

[shiyu1994]

@jameslamb Thanks. Sorry for missing that...

[jameslamb]

No problem! Appreciate you reporting this, and sorry we didn't get the fix merged sooner.