Revisit ok_to_fail EndToEndThrottlingTest test_throttling

[nvartolomei]

Since seastar upgrade the following test began to failing often #14139

The test sets bandwidth limit for tiered storage and redpanda increments a metric every time throttling is applied. Beside throttling the network, redpanda also sets disk bandwidth rate limit for the cloud storage scheduling group. Pre-upgrade it seems that disk rate limit didn't work exactly as expected and instead the network throttling would be applied often. After the upgrade, it seems that disk rate limit is applied better and we never throttle the network, hence we never increment the metric. This results in the test failing.

Marked as ok_to_fail for now. We need to investigate how to rewrite the test to make sure that throttling works as expected. We don't want to remove the disk rate limitting.

JIRA Link: CORE-2922