khepri_cluster: Fix race condition in the reset code

[dumbbell]

Why

The khepri_cluster:reset() function is mainly used to uncluster a node. To make sure that both parties (the leaving node and the rest of the cluster) see the same cluster membership in the end, we perform two resets:

1. We remove the leaving member from a remote member (if the node is clustered).
2. We remove the leaving member from its own view of the cluster.

This way, if the leaving node was out-of-sync about the cluster membership because it lost its state for instance, we are sure that at the end, everyone agrees.

However, when the leaving node is removed using a remote member, that member will stop the leaving Ra server. Therefore, when we try the second remove on the leaving node, we might get a {error, noproc} error because the Ra process already exited.

How

We adopt the same solution as the error handling done with wait_for_leader(): if ra:remove_member() returns the noproc error, we consider that's ok and proceed with the reset.

This should fix a rare transient failure that I saw in CI but was never able to reproduce locally.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 5 lines in your changes missing coverage. Please review.

Project coverage is 89.53%. Comparing base (c5722bf) to head (8491582).
Report is 2 commits behind head on main.

Files with missing lines	Patch %	Lines
src/khepri_cluster.erl	0.00%	5 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #294      +/-   ##
==========================================
- Coverage   89.68%   89.53%   -0.15%     
==========================================
  Files          21       21              
  Lines        3188     3192       +4     
==========================================
- Hits         2859     2858       -1     
- Misses        329      334       +5     

Flag	Coverage Δ
erlang-25	88.78% <0.00%> (+0.04%)	⬆️
erlang-26	89.44% <0.00%> (+0.04%)	⬆️
erlang-27	89.47% <0.00%> (-0.12%)	⬇️
os-ubuntu-latest	89.53% <0.00%> (-0.05%)	⬇️
os-windows-latest	89.53% <0.00%> (-0.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[dumbbell]

The patch coverage is 0% because a follow-up changes to khepri_cluster will greatly increase the chance to hit this error (100% for me locally). This is how the problem was discovered. However, I wanted to commit this one first because CI will likely fail for that other patch.