Data tiering #22
Description
Now that we have a single latest index and series of archive (arc) indices implemented with a slightly different index template, we can do things like allocate arc indexes to colder tiers. This would be done using the _tier_preference setting to try and get arc indices on colder tiered and keep the latest on hotter nodes. Of course this would require us to implement this in the infrastructure as well, but it might be worth adding these tier preferences to the templates anyway even if no tiering exists in the underlying nodes. As far as I understand it, if you specify that you want something on cold tier but there aren't nodes available to handle it then it'll just go on a data node which you have to have anyway so it shouldn't break anything.