Fix bug in eventTime comparison

The current logic when collecting Selinux events for logging is as
follows:

1. The first time the LogsService run, it collects events in the
    [mLastWrite_N, mLastWrite_N+1] time bucket,
2. The next time, it collects events in the
    [mLastWrite_N+1, mLastWrite_N+2] time bucket.

By this logic, events happening at time mLastWrite_N+1 are collected two
times. With this CL, a LogService run will collect events in the (mLastWrite_N+1, mLastWrite_N+2] bucket (startpoints excluded), avoiding the double collection.

Bug: 295861450
Test: atest SelinuxAuditLogsCollectorTest SelinuxAuditLogsBuilderTest
Change-Id: I3668e40789621db29b1dea73797fd56b6fd77bed