Fix flaky rate limit integration tests

[aarongable]

We've observed flaky failures of two integration tests, both of which were touched by https://github.com/letsencrypt/boulder/pull/7666.

Failure of the rate limit test:

--- FAIL: TestCertificatesPerDomain (2.88s)
    ratelimit_test.go:57: Somehow managed to issue third certificate: expected error but received none

Failure of the otel tracing test:

did not find span boulder-wfe2::sa.StorageAuthorityReadOnly/GetRegistration with parent 'd292b31fcf42ec85'
did not find span boulder-wfe2::redis.pipeline get with parent 'c0502d9ccfbdf9f2'
did not find span boulder-wfe2::/acme/new-order with parent '13522f69f46460f8'
did not find span integration.test::TraceTest with parent ''
--- FAIL: TestTraces (13.76s)
    otel_test.go:278: Failed to find expected spans in Jaeger for trace e0c52d9859552f2cca98df72bbd2ab36

I suspect that these are related, both because they showed up at the same time shortly after #7666 landed, and because they both concern rate limits: the span that the otel trace is looking for is part of the redis ratelimit lookup code.