Add system.user_processes table to metric scraping for clickhouse versions 23.5 and newer

It allow to plot per user memory usage. It also have counters metrics (ProfileEvents), but having all metrics per user in prometheus is probably and overkill.

I would scape user memory (i.e. field memory_usage ) - it's important (especially in the context of max_memory_usage_for_user, which sometimes misbehave).

SELECT *
FROM system.user_processes

Row 1:
──────
user:              default
memory_usage:      20280
peak_memory_usage: 20312
ProfileEvents:     {'Query':999,'SelectQuery':774,'InsertQuery':11,'QueriesWithSubqueries':1050,'SelectQueriesWithSubqueries':836,'FailedQuery':10,'FailedSelectQuery':2,'FailedInsertQuery':8,'QueryTimeMicroseconds':291074819,'SelectQueryTimeMicroseconds':14543907,'InsertQueryTimeMicroseconds':90045780,'OtherQueryTimeMicroseconds':186485132,'FileOpen':1284655,'ReadBufferFromFileDescriptorRead':918300,'ReadBufferFromFileDescriptorReadBytes':1911963707,'WriteBufferFromFileDescriptorWrite':344074,'WriteBufferFromFileDescriptorWriteBytes':17420233,'FileSync':81,'FileSyncElapsedMicroseconds':665642,'ReadCompressedBytes':1836958979,'CompressedReadBufferBlocks':267007,'CompressedReadBufferBytes':18302182187,'OpenedFileCacheHits':4495,'OpenedFileCacheMisses':940660,'OpenedFileCacheMicroseconds':180234,'IOBufferAllocs':1317781,'IOBufferAllocBytes':363233870154,'ArenaAllocChunks':70594,'ArenaAllocBytes':291631104,'FunctionExecute':119922,'TableFunctionExecute':56,'MarkCacheHits':25235,'MarkCacheMisses':1084,'CreatedReadBufferOrdinary':945155,'DiskReadElapsedMicroseconds':58717702,'DiskWriteElapsedMicroseconds':1530726,'NetworkReceiveElapsedMicroseconds':26602,'NetworkSendElapsedMicroseconds':151059,'NetworkReceiveBytes':123182,'NetworkSendBytes':4906096,'InsertedRows':11,'InsertedBytes':791,'DistributedConnectionTries':74,'DistributedConnectionUsable':72,'DistributedConnectionFailTry':2,'SuspendSendingQueryToShard':75,'CompileFunction':3,'CompileExpressionsMicroseconds':577507,'CompileExpressionsBytes':24576,'SelectedParts':8376,'SelectedRanges':8376,'SelectedMarks':461023,'SelectedRows':2885110301,'SelectedBytes':21744060606,'WaitMarksLoadMicroseconds':1669157,'LoadedMarksCount':70572,'LoadedMarksMemoryBytes':301288,'ContextLock':46795,'ContextLockWaitMicroseconds':5312,'RWLockAcquiredReadLocks':2283,'RWLockReadersWaitMilliseconds':3,'PartsLockHoldMicroseconds':16376,'RealTimeMicroseconds':401724887,'UserTimeMicroseconds':48506195,'SystemTimeMicroseconds':23116288,'SoftPageFaults':71168,'HardPageFaults':1362,'OSIOWaitMicroseconds':148370000,'OSCPUWaitMicroseconds':36175758,'OSCPUVirtualTimeMicroseconds':71583568,'OSReadBytes':890511360,'OSWriteBytes':1409347584,'OSReadChars':1946936584,'OSWriteChars':17517544,'CreatedHTTPConnections':204,'CannotWriteToWriteBufferDiscard':567,'QueryProfilerRuns':467,'S3WriteMicroseconds':5838285,'S3WriteRequestsCount':16,'DiskS3WriteMicroseconds':5838285,'DiskS3WriteRequestsCount':16,'S3DeleteObjects':16,'DiskS3DeleteObjects':16,'EngineFileLikeReadFiles':91,'ReadWriteBufferFromHTTPPreservedSessions':192,'ThreadPoolReaderPageCacheMiss':54656,'ThreadPoolReaderPageCacheMissBytes':1865069139,'ThreadPoolReaderPageCacheMissElapsedMicroseconds':3659429,'AsynchronousReadWaitMicroseconds':26165498,'LogInfo':16,'LogWarning':2,'LogError':20}

Row 2:
──────
user:              admin
memory_usage:      145144302563
peak_memory_usage: 145587752989

Start by locating the Go metric-scraping entry point and the ClickHouse version checks. Verify how system tables and scraped fields are represented, then add memory_usage from system.user_processes for ClickHouse 23.5 and newer without exporting the ProfileEvents counters. Done means per-user memory metrics are exposed and existing tests or validation pass.