feat: switch solver to index-based representation with optimized first-guess strategy

CLAUDE.md

@@ -2,7 +2,8 @@
 
 Mastermind solver using Java algorithms (performance) + Python UI (terminal).
 Goal: Efficiently solve c=9, d=9 cases.
-Status: Rewriting codebase; currently focused on Java algorithm only.
+Status: Java algorithm complete and performing well (~2s for a full 9x9 solve). Currently in cleanup phase (comments,
+tests, chores). Python UI not yet started.
 
 ### Code Organization
 
@@ -20,13 +21,14 @@ Status: Rewriting codebase; currently focused on Java algorithm only.
 5. `GuessStrategy.select()` — chooses which guesses and secrets arrays to pass into `BestGuess`
 6. `MastermindSession` — manages a full game: history, solution space, strategy-based suggestions, undo
 
-### Next Move / Current Move
+### Current Focus
 
-- Run a 9x9 demo and profile the code to determine where the bottleneck is.
-- Continue micro-optimizing code to increase efficiency
+- Java side mostly done; touching up code quality, expanding test coverage, misc chores.
+- Next major phase: Python UI.
 
 ### Preference
 
 - Stick to primitive type unless there is a reason not to.
 - Unless necessary, do not write extra class and objects. Be simple.
-- Do not run any tests or benchmark for me unless specifically instructed.
+- Do not run any tests or benchmark for me unless specifically instructed.
+- DO NOT TOUCH the average performance in benchmarks. Don't delete, don't modify, don't change, don't update.