[FIX] CassandraMailRepository should be more resilient to extra large…

[chibenwa]

… mail repository

Symptoms

{
 "additionalInformation": {
   "timestamp": "2026-08-01T03:51:44.221491809Z",
   "type": "ExpireMailboxTask",
   "runningOptions": {
     "usersPerSecond": 1,
     "mailbox": "Trash",
     "useSavedDate": true,
     "user": null,
     "byExpiresHeader": false,
     "olderThan": "30d"
   },
   "mailboxesExpired": 2602,
   "mailboxesFailed": 446,
   "mailboxesProcessed": 117461,
   "messagesDeleted": 21302
 },
 "submittedFrom": "james10_pub.sante-ra.fr",
 "executedOn": "james21_pub.sante-ra.fr",
 "cancelledFrom": null,
 "submitDate": "2026-07-30T19:01:56.770+0000",
 "type": "ExpireMailboxTask",
 "status": "failed",
 "taskId": "62ab7ee7-951a-4da2-af40-3d0c78b97aa2",
 "startedDate": "2026-07-30T19:01:56.786+0000",
 "failedDate": "2026-08-01T03:51:44.243+0000"
}

Subsequent runs fails with

reactor.core.Exceptions$ErrorCallbackNotImplemented: org.apache.james.backends.cassandra.utils.CassandraAsyncExecutor$CassandraAsyncExecutorException: Failed executing SELECT count(*) AS count FROM mailrepositorykeys WHERE name=:name
Caused by: org.apache.james.backends.cassandra.utils.CassandraAsyncExecutor$CassandraAsyncExecutorException: Failed executing SELECT count(*) AS count FROM mailrepositorykeys WHERE name=:name
at org.apache.james.backends.cassandra.utils.CassandraAsyncExecutor.lambda$withStatement$0(CassandraAsyncExecutor.java:75)

Diagnostics

Classic tumbstone issue: 100.000 threashold is exceeded.