Implement Solver Pattern abstraction

[Hugo-W]

Implement Solver Pattern abstraction

Description

After solver consolidation, refactor solvers into a proper abstraction hierarchy to improve maintainability and extensibility.

Tasks

• Create abstract Solver base class in pyeeg/solvers.py
• Refactor existing regression methods as concrete Solver subclasses:
  • SVDSolver
  • LSTQSolver
  • ConjugateGradientSolver
  • Others as appropriate
• Update TRFEstimator to accept a solver instance via dependency injection
• Ensure all solvers implement consistent interface

Acceptance Criteria

• Abstract Solver base class with clear interface
• All existing solvers are concrete implementations
• TRFEstimator can use any solver via composition
• Easy to add new solver types
• All solvers produce consistent results

Priority

Medium

Dependencies

• Issue Clean README.md naming and references #4 (Consolidate duplicate _svd_regress) must be completed first
• Issue Audit docs/source/ for naming consistency #5 (Move root solver.py into pyeeg/solvers.py) must be completed first

Epic

Refactoring for Modularity

Related

Builds on solver architecture work from NEXT_STEPS.md Step 3