This pull request introduces several enhancements to the project, including the addition of a continuous integration (CI) workflow, dependency updates, improved documentation, and significant refactoring of two key classes (AlphaOrder and BetaToAlpha), as well as a general cleanup of all the classes in the package (consensusUnion, consensusBES, etc). These changes aim to improve code maintainability, testing capabilities, and overall project structure.

CI and Build System Improvements:

• Added a GitHub Actions workflow (.github/workflows/ci.yml) to run Maven tests on pushes and pull requests targeting the main branch. This includes Java setup, Maven package caching, and test execution.
• Updated the pom.xml to include JUnit 5 (junit-jupiter) and Apache Commons Math3 dependencies, upgraded the Maven Surefire Plugin to version 3.1.2, and added the JaCoCo plugin for code coverage reporting. [1] [2]
• Created unit tests to validate the functionality of the whole project. The coverage is shown in a badge in README.md

Documentation Enhancements:

• Updated the README.md to include a CI status badge linking to the new GitHub Actions workflow.

Refactoring and Code Improvements:

AlphaOrder class:

• Refactored the class to improve readability and maintainability, including adding comprehensive Javadoc comments and replacing legacy heuristic methods with a single computeAlpha method for calculating ancestral node order.
• Introduced helper methods (checkExceptions, selectChild, coverEdge) to modularize logic and simplify the main computation flow.

BetaToAlpha class:

• Enhanced the class with detailed Javadoc comments, encapsulated fields, and added a method to compute a random alpha order for testing. [1] [2]

BackwardEquivalenceSearchDSep class:

• This class extracts the $$BES_d$$ component of the consensusBES algorithm so that it can be easily readable and maintained.
• This class is enhanced with a cache that now uses DSeparationKey to avoid repeating calculations.
• Like previous implementations, it uses the d-separation calculation, which is now in the Utils class.

ConsensusUnion class:

• Cleaned up the class to avoid repeating code.

ConsensusBES class:

• Separated the fusion method into two parts: First, a ConsensusUnion object that performs the consensus, and then a BackwardEquivalenceSearchDSep that applies the $$BES_d$$ algorithm to obtain the optimum fusion.