Add lint: bevy::insert_event_resource

[BD103]

This is rebased off of #84, so that will have to be merged first. Closes #40.

This checks for cases where App::init_resource::<Events<T>>() and friends are used instead of App::add_event::<T>(). I used the following code to test it:

use bevy::prelude::*;

#[derive(Event)]
struct MyEvent;

fn main() {
    App::new()
        .insert_resource(foo())
        .init_resource::<Events<MyEvent>>();
}

fn foo() -> Events<MyEvent> {
    Default::default()
}

Put it in bevy_lint/examples/lint_test.rs, cd into bevy_lint, then run cargo build && cargo run -- --example lint_test. The output should look like this:

error: called `App::init_resource::<Events<T>>()` instead of `App::add_event::<T>()`
 --> bevy_lint/examples/lint_test.rs:9:10
  |
9 |         .init_resource::<Events<MyEvent>>();
  |          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  |
  = note: `#[deny(bevy::insert_event_resource)]` on by default
help: inserting an `Events` resource does not fully setup that event
  |
9 |         .add_event::<MyEvent>();
  |          ~~~~~~~~~~~~~~~~~~~~~~

error: called `App::insert_resource(Events<T>)` instead of `App::add_event::<T>()`
 --> bevy_lint/examples/lint_test.rs:8:10
  |
8 |         .insert_resource(foo())
  |          ^^^^^^^^^^^^^^^^^^^^^^
  |
help: inserting an `Events` resource does not fully setup that event
  |
8 |         .add_event::<MyEvent>()
  |          ~~~~~~~~~~~~~~~~~~~~~~

error: could not compile `bevy_lint` (example "lint_test") due to 2 previous errors

[janhohenheim]

Lint is robust to my testing. Treat all of my comments on the PR as very minor opinions, I'm fine either way :)

[BD103]

I'm going to let this sit for a few hours in case anyone has some last-minute feedback. After that, it's getting merged!