Make type.__new__() raise clear errors instead of returning None

[emmanuel-ferdman]

Type of Changes

	Type
✓	🔨 Refactoring

Description

This PR improves error reporting for type.__new__() validation by replacing silent None returns with descriptive InferenceError exceptions. Previously, when type.__new__() received invalid arguments (wrong metaclass type, invalid name, bad bases, etc.), the code would return None without explanation. Now it raises InferenceError with clear messages like "type.new() requires a class for metaclass" or "type.new() requires 4 arguments, got 3". This makes debugging easier and brings the validation in line with other astroid error handling patterns. All existing tests pass without modification since they already expected InferenceError for these cases.

This is part of the on-going work in PR #2847.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.37%. Comparing base (019e2b2) to head (1688ef9).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2851   +/-   ##
=======================================
  Coverage   93.37%   93.37%           
=======================================
  Files          92       92           
  Lines       11148    11150    +2     
=======================================
+ Hits        10409    10411    +2     
  Misses        739      739           

Flag	Coverage Δ
linux	93.23% <100.00%> (+<0.01%)	⬆️
pypy	93.37% <100.00%> (+<0.01%)	⬆️
windows	93.35% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
astroid/bases.py	89.69% <100.00%> (+0.05%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[DanielNoord]

Is there a way we can test this?

I do think this deserves a Changelog entry

[emmanuel-ferdman]

@DanielNoord The existing tests (test_type__new__invalid_mcs_argument, test_type__new__invalid_name, test_type__new__invalid_bases) cover these changes. While we added descriptive error messages in the code for developer clarity, they (at least currently) don't propagate to end users because Call._infer() catches all InferenceError exceptions. Even pylint currently siliently ignores the InferenceError error. The important improvement is that we now raise errors immediately instead of returning None and falling through to potentially incorrect inference paths.

I have added the changelog, hopefully in the correct place 😄