fix: check only first sublevel for validators with each_item

[PrettyWood]

Change Summary

When using a validator with each_item, the items are all validated
one by one. But if an item is also iterable the subitems are then
validated because the validator is kept as it is.
Now the validator passed to the items is changed and won't be propagated

Related issue number

closes #1255
closes #1933

Checklist

• Unit tests for the changes exist
• Tests pass on CI and coverage remains at 100%
• Documentation reflects the changes where applicable
• changes/<pull request or issue id>-<github username>.md file added describing change
  (see changes/README.md for details)

[codecov]

Codecov Report

Merging #1991 (4fad457) into master (13a5c7d) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master     #1991   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           21        21           
  Lines         4199      4202    +3     
  Branches       854       855    +1     
=========================================
+ Hits          4199      4202    +3     

Impacted Files	Coverage Δ
pydantic/fields.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 13a5c7d...4fad457. Read the comment docs.

[samuelcolvin]

I'll reply on the issue.

[samuelcolvin]

see #1933 (comment), let's update this to add recursive.

[PrettyWood]

@samuelcolvin Glad you accepted the proposal! I added the recursive kwarg to ensure the breaking change is easy to fix. But as you can see in my test the behaviour before was really weird: it was not really recursive but more a validation for the "last sublevel" based on the field type!.
Indeed

things: List[Tuple[str, str]]

would trigger the validation for each field of each tuple BUT

things: List[tuple]

would trigger the validation for each tuple (like the new behaviour)

So I think the recursive kwarg is wrong in fact sorry. I'm more in favor of

• actually drop it and make a real breaking change (but again I think it was a bug)
• add a kwarg but I would rename recursive to something like last_sublevel or deep or whatever and explain how it works (as the behaviour is really weird)

WDYT?

[samuelcolvin]

Sorry for the slow reply.

I agree recusive is wrong. Let's drop it and call this a breaking change.

[PrettyWood]

Great glad we agree then! :) I removed the last few commits and just added the "breaking change" note with some basic explanation