fix: resolve ignored individual NUTS level configurations

[stan-buren]

Fixes #2207. Individual country overrides in clustering.administrative.countries were silently ignored because the script filtered admin_levels at the root level instead of parsing the nested 'countries' dictionary.

• Update dict comprehension in base_network.py and cluster_network.py to extract country levels from nested 'countries' config
• Add unit and regression tests covering overrides extraction, key leaks, and edge cases
• Update release notes

Closes #2207.

Changes proposed in this Pull Request

Problem & Root Cause

PyPSA-Eur allows users to define custom NUTS levels for specific countries (e.g. countries: {'DE': 2}) in their configuration file under clustering.administrative.countries. During configuration loading and validation, this config is parsed into a dict (admin_levels) containing the keys "level" and "countries", where "countries" holds the nested dictionary with per-country levels.

Previously, scripts/base_network.py and scripts/cluster_network.py parsed this using:

country_level = {
    k: v for k, v in admin_levels.items() if (k != "level") and (k in countries)
}

When iterating over admin_levels.items(), the key k is either "level" or "countries". The condition (k in countries) evaluates whether the string "countries" matches any of the active country codes (e.g., ["DE", "FR"]), which always returns False. As a result, the country_level dict was silently created as an empty dict {} and all user-defined per-country overrides were silently ignored.

Solution

1. Config parsing fix: Updated the dictionary comprehension in both scripts/base_network.py and scripts/cluster_network.py to:

   country_level = {
       k: v for k, v in admin_levels.get("countries", {}).items() if k in countries
   }

   This targets the nested "countries" dictionary and filters keys based on the active countries list.
2. Added 6 unit and regression tests in test/test_build_admin_shapes.py, including a regression test that explicitly demonstrates the old code's silent failure.

All 6 unit tests pass successfully locally.

Checklist

Required:

• Changes are tested locally and behave as expected.
• Code and workflow changes are documented.
• A release note entry is added to doc/release_notes.md.

If applicable:

• Changes in configuration options are reflected in scripts/lib/validation.
• For new data sources or versions, these instructions have been followed.
• New rules are documented in the appropriate doc/*.md files.

[stan-buren]

I've extracted the country NUTS level extraction logic into a shared helper function extract_country_level in _helpers.py, and refactored both base_network.py and cluster_network.py to use it.

I also updated the tests to import this helper directly, removed the local duplicate definition, and cleaned up the redundant regression test that contained a copy of the old buggy comprehension.