Unexpected css parsing caused by the default prefix : `descendant-or-self::`

Description

This is my first time submitting an issue, so please forgive me if there are any problems. Hoping it helps.

When I use the css selector :nth-child(2) in such an html code structure, I encountered an unexpected code behavior:

<div>
    <div></div>
    <div>(1)Base this div (3)But I actually got this one
        <div></div>
        <div>(2)I want to get this div</div>
        <div></div>
        <div></div>
        <div></div>
        <div></div>
    </div>
</div>

Steps to Reproduce

You can reproduce 100% of the phenomenon I described with the following simple code.

import scrapy
import requests

response = requests.get('https://shouji.baidu.com/detail/3981189')
selector = scrapy.Selector(text=response.text)

base_info_div = selector.css('.app-base-info-content')

print(base_info_div.css('div:nth-child(1) span:nth-child(2)::text').extract_first())
print(base_info_div.css('div:nth-child(2) span:nth-child(2)::text').extract_first())
print(base_info_div.css('div:nth-child(3) span:nth-child(2)::text').extract_first())

Expected behavior:

8.8MB
2.2.3
2020-09-25 15:50:38

Actual behavior:

8.8MB
8.8MB
2020-09-25 15:50:38

Versions

Scrapy       : 2.8.0
lxml         : 4.9.1.0
libxml2      : 2.9.14
cssselect    : 1.0.0
parsel       : 1.5.0
w3lib        : 1.21.0
Twisted      : 22.2.0
Python       : 3.10.9 (main, Mar  1 2023, 12:20:14) [Clang 14.0.6 ]
pyOpenSSL    : 21.0.0 (OpenSSL 1.1.1m  14 Dec 2021)
cryptography : 36.0.1
Platform     : macOS-13.4-arm64-arm-64bit

Additional context

After tracing the code execution, I found that it was the default prefix in the following code causing my problem.

https://github.com/scrapy/parsel/blob/8c39b2fa564295247bc46cd30e313dc449889b12/parsel/csstranslator.py#L147-L152

The prefix descendant-or-self:: causes the code base_info_div.css('div:nth-child(2)) actually select base_info_div itself firstly.

I've figured out what this prefix does and I'm wondering what the point of the default prefix is, whether it's a feature or a way to deal with some special case or just a bug?

scrapy / parsel