[ramondalmau]

Replaces Dependabot's grouped update #53 (ruff 0.17.1, pycontrails up to 0.64), which failed CI on lint and on 13 PS tests.

ruff 0.17.1

• RUF068: removes the duplicate mean and sqrt entries from the backend's __all__.
• Markdown formatting: ruff 0.17 formats Python code blocks in Markdown, so the README and docs examples now follow the same style as the code.

pycontrails 0.64: engine deterioration (a real model change)

pycontrails 0.64 changed where engine deterioration is applied:

• Before (0.60.4): the factor was folded into the tabulated eta_1 at load time, as eta_1 * (1 - 0.025).
• Now (0.64.0): the table holds the raw eta_1, and PSFlight applies eta_1 / (1 + f) when it evaluates.

openleaf copies eta_1 from those tables. Bumping the version alone would therefore have silently dropped the deterioration, and every PS fuel flow would have come out 2.5 % low. The 13 failing tests were comparisons against pycontrails catching exactly that.

The fix mirrors pycontrails:

• PSEngineData.engine_deterioration_factor defaults to 0.025, as in pycontrails.
• PSModel(..., engine_deterioration_factor=...) takes it as an option, alongside thrust_buffer.
• ps_physics.overall_propulsion_efficiency applies it.

As a result, fuel flow is exactly (1 + f) × the undeteriorated value, and the new test pins this. PS efficiencies move by 0.06 % against the old (1 - f) form; for example, A320 cruise η goes from 0.30620 to 0.30639.

pycontrails is now >=0.64,<0.65: older releases ship the pre-reduced eta_1, which would count the deterioration twice. The PS aircraft tables are byte-identical between 0.60.4 and 0.64.0, so nothing else changes.

Verification

All CI gates pass locally:

• ruff check and format
• mypy for both packages
• pytest: 1014 passed, 233 BADA-data skips; also passes with -W error::FutureWarning
• pip-audit
• mkdocs build --strict

Supersedes #53.

🤖 Generated with Claude Code