And increasing the polling period, so that first queued read is unlikely
to complete before we queued the second one.

This was meant to test that all pending reads are satisfied at once
after a poll, but it causes some spurious failures as they aren't
satisfied atomically (especially on Wine). As there's an increasing seq
counter in each poll it should be enough to compare the report data.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>