WAL retains a record of ra member after ra member has been deleted. #565
Description
Describe the bug
The WAL process keeps a small record of UId => {in_seq, ra_index()} for every process that has written anything. This is only truncated when the process is restarted and a member has no entries in the current set of wal files.
There is no way to reclaim the memory used during run time. This typically wont be a problem where ra clusters are long running but could be a problem where there is higher churn.
Reproduction steps
...
Expected behavior
.
Additional context
No response