fix: Fix flaky blocklist test

[yizzlez]

Which problem is this PR solving?

This is re: #45 (comment)

The original code has a rare race condition:

1. On line 180, the currentIndex is loaded
2. Context switch to where currentIndex is incremented
3. IncrementKey is called on the updated index
4. AggregateCounts is called on an old currentIndex. This is undefined behavior, as currentIndex is assumed to be monotonically increasing.

To validate that this race condition was fixed, I wrote a script that ran the test 100 times.

With the previous code, I observed 7 failures out of 100 runs.

With the new code, I observed 0 failures out of 100 runs.

Short description of the changes

Small change to unit tests.

[kentquirk]

Thank you for this!