test(e2e): add Behave step definitions for context merging

[balgaly]

Fixes #500

Per @gruebel's note on #565 — the actual fix for #500 is the missing E2E tests, not a new setter, since Python idiomatically uses direct attribute access. The merging precedence in the SDK is already correct (openfeature/client.py:422-429); this PR proves it with Gherkin scenarios from the spec.

Summary

• Bump the spec submodule to 130df3eb so contextMerging.feature is picked up by the existing poe e2e task.
• Add tests/features/environment.py with a before_scenario hook that resets provider, hook, API context, and transaction context state. This prevents cross-scenario state leaks — without it, contextMerging scenarios can pollute the following evaluation.feature run.
• Add tests/features/steps/context_merging_steps.py:
  • RetrievableContextProvider: captures the merged EvaluationContext it receives so assertions can inspect what the SDK passed through.
  • Step definitions covering every scenario in contextMerging.feature: single-level insert, multi-level insert, and per-key overwrite precedence across API / Transaction / Client / Invocation / Before Hooks.

Client-level context is set via direct attribute assignment on OpenFeatureClient.context rather than a new setter. This matches the Pythonic pattern called out in #565 review and requires no API surface change.

Test plan

• poe e2e (with PYTHONIOENCODING=utf-8 on Windows): 4 features / 50 scenarios / 233 steps passing
• poe test: 155 pytest tests passing
• ruff check + ruff format clean

Supersedes

Closes #590 in favor of this approach.

[gemini-code-assist]

Code Review

This pull request introduces BDD test steps and environment setup for testing context merging logic in the OpenFeature Python SDK. It includes a custom provider to capture merged contexts and steps to populate context at various precedence levels (API, Transaction, Client, Invocation, and Hooks). Feedback focuses on correctly handling the "targeting_key" as a first-class property rather than a generic attribute during both context population and assertion, suggesting the use of the SDK's built-in merge method for consistency.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.33%. Comparing base (7d9229f) to head (a4abc37).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #593   +/-   ##
=======================================
  Coverage   98.33%   98.33%           
=======================================
  Files          45       45           
  Lines        2157     2157           
=======================================
  Hits         2121     2121           
  Misses         36       36           

Flag	Coverage Δ
unittests	98.33% <ø> (ø)

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

☔ View full report in Codecov by Sentry.
📢 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.

[gruebel]

thanks for the work 🍻 overall looks good, but I added a few comments

[toddbaert]

Hey @balgaly ... will you be picking this back up? Altnernatively, @karnyadavdev may want to take it over since they opened something similar as a duplicate.

[karnyadavdev]

@toddbaert, sounds good. I can take this over and push the remaining fixes