dotNext: leader memory leak when a node is down in linux - raspberry or debian

Reproduced it also with the latest in main 14.3.

It also reproduces in linux - debian wsl.

To reproduce:

• modify the DataModifier of the RaftNode example so that it saves every 0.1 seconds (10 times x second)
• for x64 run (also reproduces in arm devices): dotnet publish Examples\RaftNode -p:PublishProfile=FolderProfile -c Release -r linux-x64 --self-contained -f net6.0
• run the 3 nodes as usual
• a few mins later kill one of the non leader nodes
• monitor the memory and confirm a few minutes later the leader has gained significant more memory
• if you leave it running its memory usage keeps growing / I have had a working set of well over 1GB