Evaluate merging nosedrum's application command logic

[jchristgit]

With nostrum having functionality now that is pretty exquisite but most people might not end up using (multi-node distribution as an example), it feels appropriate to also improve the user experience for the common use cases.

nosedrum currently consists of four parts:

• Application command support: command behaviour, registering commands, dispatching them, handling components
• Message cache: ETS-based message cache
• Text command support: my bot is extensively built on it but I consider it legacy. I don't think we should add it to nostrum.
• Miscellaneous helpers, mainly for text commands (predicates, converters)

I'd like to propose merging the application command support into nostrum. It's < 800 LOC, well tested, well documented, and I think it would greatly simplify common nostrum usecases.

Woudl be happy for other thoughts.

[Th3-M4jor]

I'm for the idea, but I think we should go for something closer to what's described in #668 that you're expected to use instead of the way Nosedrum works.

It'd allow us to do a lot of things at compile time.

[jchristgit]

What's the benefit of doing these things at compile time though? I understand where you come from, but I want to avoid some form of macro hell where nobody understands what is going on.

[Th3-M4jor]

The main benefit is runtime performance, there's a fair bit of overhead with dispatch that just goes away if we're able to have it be a single :select_val followed by a static function call as opposed to dynamic dispatch.