refactor: improve physical simulation stability with RK2 integration and update documentation across the environment and task modules

feat: add Apache 2.0 license and update environment failure thresholds, recovery metrics, and success scoring.

refactor: adjust task scoring weights, update dynamics coefficients, and include score in episode logging

refactor: decouple inference logic to support multi-task execution and update success threshold to 0.80

refactor: standardize score clamping to [0.01, 0.99], update task definitions, and add health check endpoint

fix(constants): update default episode id and enable dotenv override

refactor(server): migrate FastAPI app to server module and enhance robustness

feat(graders): implement task graders with metrics and registry integration

refactor(core): integrate modular task system for episode management

feat(api): enhance endpoint validation, thread safety, and testing infrastructure

feat: implement phase 2 with non-linear transitions and robust parser

Merge branch 'main' into p1A

updated transition logic, completed parsing update and fixed inference values

teammate B's work

initialised inference and parsing

state variables and update equations and state transitions created

test suite initialised

structure initialised