Flaky test: tests/snuba/rules/conditions/test_event_frequency.py::EventFrequencyPercentConditionQueryTest::test_batch_query_percent

Failing test: `tests/snuba/rules/conditions/test_event_frequency.py::EventFrequencyPercentConditionQueryTest::test_batch_query_percent`
Sentry sha: ca860f9c94028d730c6c41f914221b3bb1f4059c
Run: https://github.com/getsentry/sentry/actions/runs/22268329341

```
tests/snuba/rules/conditions/test_event_frequency.py:511: in test_batch_query_percent
assert batch_query == {
E assert {1: 25.0, 2: 25.0, 4: 0} == {1: 20, 2: 20, 4: 0}
E 
E Omitting 1 identical items, use -vv to show
E Differing items:
E {1: 25.0} != {1: 20}
E {2: 25.0} != {2: 20}
E 
E Full diff:
E {
E - 1: 20,
E + 1: 25.0,
E ? ++
E - 2: 20,
E + 2: 25.0,
E ? ++
E 4: 0,
E }
```

Start with tests/snuba/rules/conditions/test_event_frequency.py::EventFrequencyPercentConditionQueryTest::test_batch_query_percent and run the test, using the linked GitHub Actions run as context. Review why the batch query produces 25.0 instead of the expected 20 for keys 1 and 2; done means the flaky behavior is addressed and the test passes consistently.

Written by the indexing model from the issue text.