Refactor data stream lifecycle to use the template paradigm

[gmarouli]

In this PR we migrate the DataStreamLifecycle configuration to the new "template" framework we introduced in #117357. Effectively, we split the code that is used by the data stream and the templates. This is better for the following reasons:

• The explicit nulls are scoped only in the template code and will not leak to the data stream (any more).
• The explicit nulls are configured in a predicable way using the ResettableValue and can be composed easier as well.
• In a follow up PR when it's time to incorporate the lifecycle under the failure store, it will be relatively easy since the failure store config already uses this template framework.
• We removed dedicated wrapper classes that where necessary to represent the explicit nulls.

[elasticsearchmachine]

Pinging @elastic/es-data-management (Team:Data Management)

[gmarouli]

I will work on the CI in the meantime, but I wanted to open it already for a review because it's quite a big.

[gmarouli]

I have been going back and forth about this one, should we keep it boolean or should we convert it to Boolean.

When it comes to failure store we had a specific use case where we enable it via a cluster setting. Here, we do not have such a use case yet, so I am leaning towards keep it as is. Any thoughts?

cc @dakrone