cloud_storage: insert sleep in aborted transactions

[LenaAn]

Cover letter

Reproducing #5613

[jcsp]

(Remember that #5613 was happening on clustered ducktape -- running the regular PR CI isn't going to hit the same tests)

[LenaAn]

With this sleep(10s) I got crash every time for EndToEndShadowIndexingTest locally, so I'm expecting EndToEndShadowIndexingTest to crash in CI here

[LenaAn]

so with this sleep(10s) inserted I get NodeCrash in

• ShadowIndexingWhileBusyTest.test_create_or_delete_topics_while_busy
• EndToEndTopicRecovery.test_restore.message_size=10000.num_messages=100000.recovery_overrides=.retention.bytes.1024
• EndToEndShadowIndexingTestWithDisruptions.test_write_with_node_failures
• EndToEndShadowIndexingTest.test_write
• ShadowIndexingCacheSpaceLeakTest.test_si_cache.message_size=10000.num_messages=100000.num_read=1000.concurrency=2
• ShadowIndexingTxTest.test_shadow_indexing_aborted_txs

so the crash is reproduced.

will use stable iterator with the same sleep(10s) and will see if it fixes the problem