feat(client): add request rate limiting

[gijzelaerr]

Summary

• add opt-in per-client request limiting at the outbound S7 PDU boundary
• support evenly spaced fixed-rate and burst-capable token-bucket algorithms
• support blocking and immediate rejection when the limit is reached
• cover both synchronous and asynchronous clients
• document configuration and add deterministic limiter tests

A multi-variable PDU counts as one request, while chunked and multi-packet operations count each actual PDU. Model-specific defaults are intentionally not inferred: the limiter is disabled by default and users choose a rate appropriate for their PLC and workload.

Silent drops are unsafe at the PDU boundary because the surrounding client operation would wait for a response to a request that was never sent. The non-blocking behavior therefore raises S7RateLimitError immediately.

The branch includes current master through a merge commit.

Validation

• uv run --frozen pre-commit run --all-files
• uv run --frozen --extra test --extra s7commplus pytest -q (1,902 passed, 78 skipped)
• focused limiter/client tests (163 passed, 1 skipped)
• uv build --no-sources

Fixes #628

[gijzelaerr]

@NiklasReisser Given your recent work around server behavior and application-facing semantics, could you review the request-rate-limiting API and concurrency behavior here?

[gijzelaerr]

@amorelettronico you previously tested the pure-Python client on CPU 414-5H and CPU416-2 hardware. Could you run a small real-PLC smoke test of the opt-in rate limiter on an S7-400 when convenient?\n\nThe useful check would be a low fixed rate (for example 5 requests/second), confirming reads still succeed and are spaced as configured, followed by disabling the limiter and confirming normal behavior returns. No production load test or sensitive PLC data is needed.

[amorelettronico]

Hi @gijzelaerr

Hardware smoke test — S7-416-2 DP

Tested PR #823 against a Siemens S7-416-2 DP using DB15 as a dedicated test DB.

Configuration

• Test DB: DB15
• Rate limit: 5 requests/sec
• Expected interval: 200 ms
• Algorithm: fixed
• Behavior: block

DB read

• Successfully read DB15.
• Returned values matched the expected test data.

Rate limiter — 5 requests/sec

• 15 consecutive DB reads completed successfully.
• Measured intervals:

0.200, 0.201, 0.199, 0.199, 0.202,
0.198, 0.199, 0.202, 0.201, 0.200,
0.198, 0.200, 0.201, 0.200 sec

• Average interval: 0.200 sec

Limiter disabled

• With max_requests_per_second=0, reads completed successfully without the intentional 200 ms delay.
• Measured intervals were approximately 6–15 ms, confirming that the rate limiter was no longer imposing the 5 req/sec limit.

Conclusion

The rate limiter behaves as expected on the S7-416-2 DP with a real PLC connection.

PR #823 smoke test: PASS.

[gijzelaerr]

Thanks @amorelettronico — this covers the requested hardware smoke test: successful reads, approximately 200 ms spacing at 5 requests/sec, and normal timing with the limiter disabled.

@NiklasReisser the review findings were addressed in cd631ed and replied to inline. All current checks pass, and the S7-416-2 DP hardware test now passes too. Could you take another look and submit a review? The existing review request is still pending.