Behaviour on setting TrailingLogs to 0

https://gist.github.com/fahadullah/e49a4578094f4efd7746

If TrailingLogs is set to 0 and raft node is restarted right after taking a snapshot, it tries to look up the last index but all entries were already compacted. TrailingLogs being 0 should be a valid config though a little inefficient under certain cases.

Start with the linked gist, then trace the snapshot and restart path for a Raft node configured with TrailingLogs set to 0. Reproduce the failure around the last-index lookup and verify that restarting immediately after a snapshot completes without trying to access compacted entries.

Written by the indexing model from the issue text.