placeholder to simulate MongoDB RS node failures before index creation

[viniarck]

This is related to https://github.com/kytos-ng/flow_manager/issues/160.

I'm mapping this placeholder to also document on this issue.

Italo has mapped to reproduce this linked issue with 5 nodes, he stopped node 3 and 5, and then when kytosd it got stuck during index creation, he was using version 2022.3.1.

I'll try to simulate it with 3 nodes (that we typically use in our dev env), and then if needed I'll use 5 nodes too.

[viniarck]

I managed to reproduce the issue with 5 nodes based on @italovalcy's reported information. Here's the summary:

• If the Replica Set doesn't have the majority of nodes available, then kytosd will exit since the hello command won't succeed (so we've got that part covered).
• If the Replica Set still have the majority available, index creation, even in background can take from seconds to minutes, it depends on how large the existing collection is, which is why MongoDB docs recommends to create them upfront. Our NApps will still autobootstrap them since it works when creating from scratch. With that said, from this point forward, any index operation should provide scripts just to network operators can run them upfront.
• To provide stronger guarantees, the controller will know shutdown if any NApp doesn't handle an exception during its setup() initialization https://github.com/kytos-ng/topology/issues/163#issuecomment-1734440153 (so, not only this also covers the DB execeptions, but also partly covers the scope of https://github.com/kytos-ng/kytos/issues/314

I'll send a PR and backport it to both 2023.1 and 2022.3 shortly.