docs: Clarify there's one extension instance per crawler

[jpmckinney]

The previous documentation suggested there was a single instance of each extension. This is not the case. For example, when running multiple spiders in the same process, the LogStats extension still works correctly, and it doesn't track stats per spider.

[codecov]

Codecov Report

Merging #5014 (caecf4e) into master (f95ebd8) will decrease coverage by 0.16%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #5014      +/-   ##
==========================================
- Coverage   88.01%   87.84%   -0.17%     
==========================================
  Files         158      158              
  Lines        9726     9726              
  Branches     1433     1433              
==========================================
- Hits         8560     8544      -16     
- Misses        911      928      +17     
+ Partials      255      254       -1     

Impacted Files	Coverage Δ
scrapy/robotstxt.py	75.30% <0.00%> (-22.23%)	⬇️
scrapy/core/downloader/__init__.py	92.48% <0.00%> (+1.50%)	⬆️

[Gallaecio]

Thanks, that is a good point!

However, I personally feel like this change could be reduced to appending “ per spider being run” to “instantiating a single instance of the extension class”, and things should be clear enough.

I also think that we should avoid mentioning the crawler at this stage, as the crawler object is something people reading these 2 sections may not yet know.

[jpmckinney]

@Gallaecio I've now made the simpler change you suggest.