Spike protection keeps triggering after planned audits

### Problem Statement

When we run a planned audit that causes a big but expected spike in errors, Sentry’s spike protection turns on (as expected) but then keeps triggering afterward during normal traffic.

This causes:

- Extra “spike protection triggered/ended” emails
- Some real production events being dropped even when volume looks normal

Can spike protection handle planned spikes (like audits or load tests) better so that:

- A one-time, intentional spike doesn’t make the system overly sensitive afterward
- We don’t lose normal production events or get noisy alerts once traffic is back to usual

### Product Area

Settings - Spike Protection

Start in Settings - Spike Protection and trace how planned audits affect later spike detection, event dropping, and notifications. Clarify the expected behavior for intentional one-time spikes versus normal traffic, then validate that normal events are retained and duplicate alerts do not continue after traffic returns to usual.

Written by the indexing model from the issue text.