fix: clamp counters at MAX_SAFE_COUNTER to prevent float64 precision loss

[wangarthur999999-star]

Summary

• Added MAX_SAFE_COUNTER = 2**53 constant — largest integer representable in IEEE 754 double precision
• Added negative value validation in increment() with clear error messages
• Added min(current + value, MAX_SAFE_COUNTER) clamping on every increment to prevent counter overflow beyond float64 safe integer
• Removed stale timestamp artifacts from the file

Implementation

Proactive increment-time clamping — the counter is checked and clamped on every increment() call rather than at snapshot time. This is safer than snapshot-time validation because it prevents the counter from entering an overflow state at all.

Test Coverage

16 tests in tests/test_counter_range_limits.py:

• TestIncrementValidation (4): positive value, default, negative rejection, negative message
• TestCounterRangeClamping (5): clamped at max, exactly at max, below max then increment, multiple increments, no wrapping
• TestCounterSnapshot (3): snapshot includes counters, plain dict types, float64 safety constant
• TestCounterEdgeCases (4): zero noop, large increment, independent counters, approach limit

All 16 tests pass.

Differs from competing PR #3948

PR #3948 validates at snapshot time. This implementation clamps at increment time — preventing overflow state entirely rather than filtering it on read.

🤖 Generated with Claude Code