Document how to handle action failures

[Gallaecio]

Resolves #118, resolves #119, resolves #120.

To do:

• Test both snippets manually, make sure they work as expected.

[codecov]

Codecov Report

Merging #157 (ea2da8f) into main (9b67ef1) will decrease coverage by 0.12%.
Report is 61 commits behind head on main.
The diff coverage is 92.20%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #157      +/-   ##
==========================================
- Coverage   98.18%   98.07%   -0.12%     
==========================================
  Files          11       12       +1     
  Lines         828      985     +157     
==========================================
+ Hits          813      966     +153     
- Misses         15       19       +4     

Files	Coverage Δ
scrapy_zyte_api/exceptions.py	100.00% <100.00%> (ø)
scrapy_zyte_api/_middlewares.py	94.59% <91.54%> (-1.96%)	⬇️

... and 5 files with indirect coverage changes

[kmike]

There is a lot of code in this section; I wonder if we should make it a part of the library itself instead.

It seems like a general theme: if action failed, should the response be considered a success or a failure? The current default is success, but it seems that making it a failure is better (with an option to get back), as it allows to detect more issues.

[Gallaecio]

deedfff does this for the middleware. Do we want to do something about the cache policy, or is it OK to keep that documentation-only?