Don't report ZK node exists to system.errors when a block was created concurrently by a different replica

[Algunenano]

Changelog category (leave one):

• Improvement

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

• Don't report ZK node exists to system.errors when a block was created concurrently by a different replica

Example (comes from a CI run):

version:            23.3.1.19
name:               KEEPER_EXCEPTION
code:               999
quantity:           1
last_error_message: Transaction failed (Node exists)
traceback:          DB::Exception::Exception(DB::Exception::MessageMasked&&, int, bool)
Coordination::Exception::Exception(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&, Coordination::Error, int)
Coordination::Exception::Exception(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&, Coordination::Error)
zkutil::KeeperMultiException::KeeperMultiException(Coordination::Error, std::__1::vector<std::__1::shared_ptr<Coordination::Request>, std::__1::allocator<std::__1::shared_ptr<Coordination::Request>>> const&, std::__1::vector<std::__1::shared_ptr<Coordination::Response>, std::__1::allocator<std::__1::shared_ptr<Coordination::Response>>> const&)
DB::ReplicatedMergeTreeSinkImpl<false>::commitPart(std::__1::shared_ptr<DB::ZooKeeperWithFaultInjection> const&, std::__1::shared_ptr<DB::IMergeTreeDataPart>&, std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char>> const&, unsigned long, bool)
DB::ReplicatedMergeTreeSinkImpl<false>::finishDelayedChunk(std::__1::shared_ptr<DB::ZooKeeperWithFaultInjection> const&)
DB::ReplicatedMergeTreeSinkImpl<false>::consume(DB::Chunk)
DB::SinkToStorage::onConsume(DB::Chunk)
DB::ExceptionKeepingTransform::work()
DB::ExecutionThreadContext::executeTask()
DB::PipelineExecutor::executeStepImpl(unsigned long, std::__1::atomic<bool>*)
ThreadPoolImpl<std::__1::thread>::worker(std::__1::__list_iterator<std::__1::thread, void*>)

Logs:

2023.02.24 09:44:39.462647 [ 271 ] {8d6ebeff-b02e-4700-83bf-2b965d241acb} <Trace> d_test_177ab302399e42f5a6f72ac44d8a9990.t_c16d3a255a7f4e4d82422c9337779ae1 (c2bed6e8-68c6-4035-b2cb-ff4ab0281d36): Renaming temporary part tmp_insert_all_2_2_0 to all_1_1_0 with tid (1, 1, 00000000-0000-0000-0000-000000000000).
2023.02.24 09:44:39.463916 [ 271 ] {8d6ebeff-b02e-4700-83bf-2b965d241acb} <Information> d_test_177ab302399e42f5a6f72ac44d8a9990.t_c16d3a255a7f4e4d82422c9337779ae1 (c2bed6e8-68c6-4035-b2cb-ff4ab0281d36) (Replicated OutputStream): Block with ID all_6014107614689632743_16975405373040374011 already exists (it was just appeared). Renaming part all_1_1_0 back to tmp_insert_all_2_2_0. Will retry write.
2023.02.24 09:44:39.463942 [ 271 ] {8d6ebeff-b02e-4700-83bf-2b965d241acb} <Debug> d_test_177ab302399e42f5a6f72ac44d8a9990.t_c16d3a255a7f4e4d82422c9337779ae1 (c2bed6e8-68c6-4035-b2cb-ff4ab0281d36): Undoing transaction. Rollbacking parts state to temporary and removing from working set: all_1_1_0.

The problems comes from using zkutil::KeeperMultiException to extract the path of the first error. Creating a zkutil::KeeperMultiException creates a KeeperException, which is a DB::Exception and thus reports to system.errors. Since the exception was created only for that operation, I've replaced by the equivalent functions instead.

Closes #32997

[den-crane]

closes #32997

[Algunenano]

closes #32997

Added to the PR description. Thanks!

I've been cleaning up these errors when I've found them. Most of them where fixed back in #38961 and #39016. There might be more, but at least those are the ones that have appeared in our CI.

[tavplubix]

Stateless tests (ubsan) [2/2] - #46636 (comment)
Stress test (asan) - #37664
Stress test (debug) - #46834
Stress test (tsan) - #37664
Stress test (ubsan) - #45372