final push

feat(grpo): enhance action response parsing by removing reasoning blocks and refining regex handling

feat(grpo): update max completion length and refine prompt handling for improved evaluation

feat(grpo): adjust training parameters and disable thinking mode for consistent action calls

feat(grpo): enhance training dynamics with new replay policies and update state steps

feat(grpo): update max sequence length and refine prompt formatting in training scripts

refactor: remove SakhaEnvWrapper class and streamline reward function in GRPO training script

feat(rubric): integrate SakhaRubric into SakhaEnvironment step/reward path

feat(rubric): add composable rubric scaffolding wrapping existing reward logic

fix(inference): align checklist compliance and structured run logs

feat(formatters): add structured output formatting system

feat(graders): implement new scoring logic

feat(env): implement workflow-based simulation

feat(models): added new action types and pending task system

fix: heuristic takeover when llm fails, step reward/penalty derived from grader, some other fixes

fix: copy README to Docker, enable web UI, and enhance documentation

initial cook