Autoscale bot instances with runtime context

Since the Load Balancer can handle N shards to M bot instances, we should take advantage of autoscaling the bot instances independently. I propose the following strategy:

• Give every bot instance a full CPU core
• Collect the following metrics:
  • Events per guild per hour (GuildEvents)
  • Events per instance per hour with their mCPU (InstanceEvents)
• Average the InstanceEvents to determine the events each instance can handle
• Every 6 hours, perform the following operation:
  • Using the average events per hour for the whole bot, determine how many instances are needed
  • Rescale the instances to the appropriate amount
  • Remap a distribution of guilds to instances, allowing for multiple instances per guild
  • Use said map to balance out the events across instances