ResponseCache has no file-level locking for concurrent access

[jtdub]

Summary

ResponseCache in cache.py reads and writes JSON cache files without any file-level locking. In multi-process scenarios (common in network automation managing many devices), two processes could race on the same cache file, producing corrupt reads or lost writes.

Current Behavior

• get() opens and reads JSON without locking
• set() opens and writes JSON without locking
• No fcntl.flock() or equivalent mechanism

Impact

• Corrupt cache reads leading to json.JSONDecodeError (caught, but cache miss)
• Lost writes where one process overwrites another's cached response
• Unlikely but possible partial reads of in-progress writes

Suggested Fix

Consider using fcntl.flock() (Unix) or a cross-platform file locking library like filelock for atomic read/write operations on cache files. Alternatively, consider using an in-memory cache (like cachetools) with an optional persistence layer.