[Durability] SyncWrites with persist_majority fails consistently after loading few batches of docs

Build: 6.5.0-3458

Scenario:

1. Create 2 node cluster
2. Create CB bucket with replicas=2
3. Load 500K docs with durability=PERSIST_TO_MAJORITY

Observations:

1. Seeing 'DurabilityAmbiguousException' for few docs like test_docs-143045, test_docs-18232, test_docs-392994, test_docs-80422, etc..
2. Memcached crashed with error #134 on node-1
3. After that seeing continuous 'RequestTimeoutException' with reason 'SHUTDOWN'
4. Using separate client tried to insert following docs,
   • test0
   • test1
   • test2
   • test3 - durability=MAJORITY
   • test4 - durability=PERSIST_ACTIVE
   • test5 - durability=PERSIST_MAJORITY
5. With above doc inserts all except test5 is inserted successfully and test5 always failed with 'DurabilityAmbiguousException' from server. Even tried to insert with timeout=1minute in SDK

TAF testcase:

epengine.basic_ops.basic_ops.test_doc_size,nodes_init=2,num_items=500000,doc_op=update,replicas=2,durability=PERSIST_TO_MAJORITY