AFAICS the problem is that some parts may be replaced with empty parts
(after #25820), and removed by the cleanup thread, due to it is empty
[1] (while it should not be deleted since it can download source part):
<details>
```
2021.08.18 20:11:22.687933 [ 341 ] {} <Trace> test_dpefxp.alter_table_1 (0758ca24-90e7-452c-8758-ca2490e7252c): Created log entry for mutation -1_115_115_0_146
...
2021.08.18 20:11:22.707609 [ 22825 ] {} <Trace> test_dpefxp.alter_table_6 (766ae414-e113-4965-b66a-e414e1137965): Executing log entry to mutate part -1_115_115_0 to -1_115_115_0_146
2021.08.18 20:11:22.707643 [ 22825 ] {} <Debug> test_dpefxp.alter_table_6 (766ae414-e113-4965-b66a-e414e1137965): Source part -1_115_115_0 for -1_115_115_0_146 is not ready; will try to fetch it instead
...
2021.08.18 20:11:22.709397 [ 333 ] {} <Trace> test_dpefxp.alter_table_6 (ReplicatedMergeTreeQueue): Not executing log entry queue-0000001579 for part -1_115_115_0 because it is covered by part -1_115_115_0_146 that is currently executing.
2021.08.18 20:11:22.718861 [ 22825 ] {} <Information> test_dpefxp.alter_table_6 (766ae414-e113-4965-b66a-e414e1137965): DB::Exception: No active replica has part -1_115_115_0_146 or covering part
...
2021.08.18 20:11:27.936829 [ 295 ] {} <Information> test_dpefxp.alter_table_6 (766ae414-e113-4965-b66a-e414e1137965): Going to replace lost part -1_115_115_0_146 with empty part
2021.08.18 20:11:27.957839 [ 295 ] {} <Information> test_dpefxp.alter_table_6 (766ae414-e113-4965-b66a-e414e1137965): Created empty part -1_115_115_0_146 instead of lost part
...
2021.08.18 20:11:28.731635 [ 257 ] {} <Trace> test_dpefxp.alter_table_6 (ReplicatedMergeTreeCleanupThread): Cleared 190 old blocks from ZooKeeper
...
2021.08.18 20:11:28.734507 [ 257 ] {} <Trace> test_dpefxp.alter_table_6 (766ae414-e113-4965-b66a-e414e1137965): Will try to insert a log entry to DROP_RANGE for part: -1_115_115_0_146
2021.08.18 20:11:28.779373 [ 22837 ] {} <Debug> test_dpefxp.alter_table_6 (766ae414-e113-4965-b66a-e414e1137965): Removed 1 parts inside -1_115_115_0_146.
...
2021.08.18 20:11:28.792600 [ 273 ] {} <Trace> test_dpefxp.alter_table_6 (766ae414-e113-4965-b66a-e414e1137965): Created log entry /clickhouse/tables/00993_system_parts_race_condition_drop_zookeeper_test_dpefxp/alter_table/log/log-0000003459 for merge -1_111_118_1
...
2021.08.18 20:11:28.910988 [ 354 ] {} <Error> test_dpefxp.alter_table_7 (ReplicatedMergeTreeQueue): Code: 49. DB::Exception: Part -1_111_118_1 intersects next part -1_115_115_0_146. It is a bug. (LOGICAL_ERROR), Stack trace (when copying this message, always include the lines below):
2021.08.18 20:11:31.282160 [ 305 ] {} <Error> test_dpefxp.alter_table_2 (ReplicatedMergeTreeQueue): Code: 49. DB::Exception: Part -1_111_118_1 intersects next part -1_115_115_0_146. It is a bug. (LOGICAL_ERROR), Stack trace (when copying this message, always include the lines below):
```
</details>
[1]: https://clickhouse-test-reports.s3.yandex.net/27752/59e3cb18f4e53c453951267b5599afeb664290d8/functional_stateless_tests_(release,_wide_parts_enabled).html
with the explicit data loading and verification.
If the function fails, the exception will re-throw upper,
cancelling the fetch in the handling of the replicated log entry,
but this event will also wake the PartCheckThread, which will
issue a re-fetch.
* "lock_acquire_timeout" controls for how long a query will continue to
acquire each lock on its argument tables
* "lock_acquire_timeout_for_background_operations" is a per-table
setting for storages of *MergeTree family