Fix[MQBC]: skip FSM events if primary goes down

[678098]

1. Add missing state transition UNKNOWN -> RST_PRIMARY -> UNKNOWN
   This updates primary in partition thread.

2. In FSM mode, update primary only from FSM.
   Keep directly calling storageManager->setPrimaryForPartition and storageManager->clearPrimaryForPartition in legacy mode.

3. Skip enqueueing an event to FSM if primaryLeaseId is outdated or there is a primary mismatch.

4. Update UT to catch new "Primary mismatch" error.

[dorjesinpo]

The event may not be a request in which case there is no need to build and send a response.

[678098]

This test was failing due to the early reject mechanism.
I have updated the expected output.
The error log from UT:

TEST /home/runner/work/blazingmq/blazingmq/src/groups/mqb/mqbc/mqbc_storagemana 
ger.t.cpp CASE 4                                                                  
20MAR2026_18:33:15.939 18823:140470192740224 WARN /home/runner/work/blazingmq/bla 
zingmq/src/groups/mqb/mqbc/mqbc_storagemanager.cpp:401 MQBC.STORAGEMANAGER        
Cluster (testCluster) Partition [0]: dropping stale event: source [W1, 3], event  
primaryLeaseId [1], current primaryLeaseId [1], event primary [[W1, 3]], current  
primary [[E2, 2]]                                                                 
Error /home/runner/work/blazingmq/blazingmq/src/groups/mqb/mqbc/mqbc_storagemanag 
er.t.cpp(593): 'message' ([ rId = 1 choice = [ status = [ category = E_REFUSED    
code = -10 message = "Primary mismatch" ] ] ]) == 'failureMessage' ([ rId = 1     
choice = [ status = [ category = E_REFUSED code = -207 message = "Not a replica"  
] ] ])    (failed)  

[kaikulimu]

A splendidly crafted PR!