Progressive Maze Solver (Fix Tests → BFS Shortest Path → Movement Constraints → Keys & Doors)

Problem: Progressive Maze Solver (AI-enabled coding / given codebase) You are given an existing maze-solver codebase with unit tests. The maze is a 2D grid: S: start E: exit/target #: wall (blocked) .: empty cell (walkable) Possibly also: a-f: keys A-F: doors (requires the corresponding lowercase key) You can move to adjacent cells (typically 4-directional; follow the provided code/tests). Each move costs 1. Part 1: Fix failing test cases Without changing the intended behavior, fix bugs in the current implementation so that all provided unit tests pass. Part 2: BFS shortest path Implement/complete a function that returns the shortest number of steps (or the shortest path, per tests) from S to E using BFS. Return -1/None if unreachable (per test contract). Part 3: Add movement-direction constraints Update move() / neighbors() to enforce additional movement constraints (as specified during the interview). Add necessary if checks to disallow invalid moves, while still computing the shortest feasible result. Part 4: Add keys & doors Extend the maze with keys and doors: Collect a key when you step on it. You may pass through a door A only if you already have key a. Extend BFS state to include the set of collected keys (commonly as a bitmask) and return the shortest steps from S to E. Constraints (typical; infer from tests) Grid size: R x C (e.g., up to 100x100) Number of keys: K <= 6 (common for bitmask) Example (illustrative) Grid: S.. ##. ..E Output: 4 Example Input 3 3 S.. ##. ..E Output 4