improve high availability -> set pod rules to run on different nodes for lnd/redis/mongodb

[nicolasburtey]

redis and mongodb run on 3 pods with redundancy. so that if one pod goes down the app continue working
with the PR GaloyMoney/blink#199, several instance of lnd will run in parrellel, with a failover mechanism.

to really favor high availability, those pods should not run on the same node.

it should be investigate whether we can have generic rules that apply on any kubernetes cluster (generic rules), or whether it's really instance dependent (less ideal)

[krtk6160]

I've opened an issue on the bitnami repo to add support for topology spread constraints. This is a beta feature in k8s 1.18.

I was also wondering if we can use pod anti-affinity to achieve what we want.

[krtk6160]

Looks like mongodb and redis helm charts now set a "soft" pod anti-affinity setting by default. I think we can close this issue.