This pull request introduces several improvements and fixes across documentation, CI workflows, package metadata, and the gg_partial_rfsrc function. The most significant change is a major enhancement to gg_partial_rfsrc, which now automatically handles time point alignment for survival forests and allows for more efficient evaluation grids. Additional updates improve documentation, add new suggested packages, and strengthen CI and coverage reporting.

Enhancements to gg_partial_rfsrc and Documentation:

• The gg_partial_rfsrc function now automatically snaps requested partial.time values to the nearest available time.interest values in survival forests, preventing prediction errors and making the function more robust. It also adds new arguments partial.time and n_eval for finer control, and uses quantile grids for continuous predictors to improve speed and smoothness. Documentation and Rd files are updated accordingly. [1] [2] [3] [4] [5] [6] [7] [8]

Continuous Integration and Coverage Improvements:

• The GitHub Actions workflows for R CMD check and standard checks are updated to handle R-devel failures gracefully (e.g., due to upstream changes like the lazyeval PREXPR removal), with steps to bootstrap rcmdcheck and allow failures on R-devel without failing the whole workflow. [1] [2] [3] [4]
• Adds a codecov.yml configuration file to enforce minimum code coverage thresholds (65% for main, 60% for PR patches) and configures reporting behavior.
• Ignores codecov.yml in R builds by updating .Rbuildignore.

Package Metadata and Documentation:

• Updates the package version to 2.7.0.9001 and the date, and adds knitr and plotly to Suggests in DESCRIPTION. [1] [2]
• Updates _pkgdown.yml to add new tutorial articles to the documentation site.

Cleanup:

• Removes the .claude/settings.local.json file, cleaning up project configuration.This pull request adds several improvements and enhancements to the gg_partial_rfsrc function and its documentation, as well as minor updates to package metadata and CI workflow configuration. The most important changes are the addition of new arguments and logic to better handle survival forests and continuous variables, improved documentation, and updates to continuous integration for better R-devel compatibility.

Enhancements to gg_partial_rfsrc:

• Added partial.time argument to gg_partial_rfsrc, allowing users to specify desired time points for survival forests. Values are automatically snapped to the nearest available time.interest value to avoid errors in randomForestSRC::partial.rfsrc. Default behavior now uses three quartile points of time.interest if partial.time is NULL. [1] [2] [3] [4] [5] [6]
• Added n_eval argument to control the number of evaluation points for continuous predictors, improving performance by evaluating on a quantile grid instead of all observed values. [1] [2] [3] [4]
• Refactored internal logic to use helper functions for evaluation grid creation and partial dependence calculation, improving code clarity and maintainability. [1] [2]

Documentation improvements:

• Expanded Roxygen and Rd documentation for gg_partial_rfsrc to describe the new arguments, survival forest handling, and best practices for logical predictors. [1] [2] [3] [4]

Continuous integration and metadata updates:

• Updated GitHub Actions workflows to allow R-devel jobs to fail without failing the whole workflow, due to potential upstream incompatibilities. [1] [2]
• Updated package version to 2.7.0.9001 and date in DESCRIPTION. Added knitr and plotly to Suggests. [1] [2]
• Added a Tutorials section to the pkgdown site configuration.
• Removed the .claude/settings.local.json file.