[ViktorBystrov72]

Identify the offending key when Dict(per_key_traits=...) rejects a value, and preserve the expected type through nested dictionaries and containers. A bad string inside a nested dictionary is no longer reported as a dictionary that must be a dictionary; errors involving a list retain the list context.

Regression cases cover the reported schema, nested dictionaries and lists, custom TraitError messages, and the existing fallback value_trait diagnostics.

Related to the closed #978, which added an outer key label while leaving nested diagnostics outside its scope. This change carries the existing TraitError context through the per-key validation path.

Fixes #920.

AI assistance: Codex assisted with the implementation and regression tests. 🤖🤖

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.25%. Comparing base (a576f24) to head (617b73e).

Components	Coverage Δ
traitlets	85.67% <100.00%> (+0.04%)	⬆️
tests	99.08% <100.00%> (+<0.01%)	⬆️

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

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