xpath selectors with position() return incorrect data in python

[Giacomo Debidda]

I am using lxml with XPath selectors to get the content of a table on this page:

https://opensky-network.org/apidoc/rest.html#response

I need to get the fields of the second table.

If I open Chrome Web tools I can get the expected results with any one of these XPath selectors:

1. $x('//div[@id="all-state-vectors"]/div[@id="response"]/div[position()=2]/table/tbody/tr')
2. $x('//div[@id="all-state-vectors"]/div[@id="response"]/div[2]/table/tbody/tr')
3. $x('//div[@id="all-state-vectors"]/div[@id="response"]/div[position()=last()]/table/tbody/tr')
4. $x('//div[@id="all-state-vectors"]//table/thead/tr/th[contains(text(), "Index")]/parent::*/parent::*/parent::*/tbody/tr')

However, only the 4th selector returns the expected results from a Python script which uses lxml.html.

### Minimal reproducible example

```python
import requests
import lxml.html

PAGE = 'https://opensky-network.org/apidoc/rest.html'
XPATH_SELECTOR = '//div[@id="all-state-vectors"]//table/thead/tr/th[contains(text(), "Index")]/parent::*/parent::*/parent::*/tbody/tr'

res = requests.get(PAGE)
root = lxml.html.fromstring(res.content)
fields = root.xpath(XPATH_SELECTOR)
assert len(fields) == 17
```

### My environment

```sh
Python : sys.version_info(major=3, minor=6, micro=5, releaselevel='final', serial=0)
lxml.etree : (4, 2, 5, 0)
libxml used : (2, 9, 8)
libxml compiled : (2, 9, 8)
libxslt used : (1, 1, 32)
libxslt compiled : (1, 1, 32)

```

[scoder (scoder)]

XPath is implemented in libxml2, not in lxml.