Make it more obvious that the user intends to build a full reactive app

[snicoll]

A recent PR made me wonder if, rather than allowing users to switch off the creation of Spring Data repositories, we should allow them to switch off the whole infrastructure, (including repositories).

I don't remember why we've introduced a boolean flag to disable repositories specifically but it moved to a type enum where users can define if they want to work with imperative or reactive. It turns out that more and more drivers are offering both mode so auto actually means both.

This also means that if you don't do anything, you get both a template and a reactive template. This can be what you want if you want to use the imperative style on startup but it would be nice if we offered a way to easily back-off.

Concretely we'd have a spring.data.<xxx>.type and selecting imperative means that the whole reactive infra would not be configured.

[snicoll]

Reviewing this a bit more, such a flag could help but that may be only one part of the story. The problem has become much more apparent now that Neo4j has imperative and reactive support in the same driver and support for transaction management.

Using Flux as a classpath check to determine if we need to configure the reactive variant of a Spring Data store (or anything that's reactive in general) is too limited, especially considering the WebClient use case where one might want to use it while being on a fully imperative app.

Users developing a reactive app are also keen to use the imperative variant on startup (to initialize or query the store in @PostConstruct or ApplicationRunner). They won't use declarative transaction management there probably.