Classify suggestions

It would be great if the output for each suggestion had a classification, for example:

```
[{
suggestion: "omit 'So' from the beginning of sentences",
type: 'so',
index: 0, offset: 2
}, {
suggestion: "'was stolen' is passive voice",
type: 'passive-voice',
index: 11, offset: 10
}]
```

At the moment, this can only be achieved by disabling all but a single check and running each individually. I'd love to use this linter to aggregate the various issues it checks for rather than implementing each linter individually. I think classifications would be helpful for such a use case.

No files, tests, or entry points are named. Start by locating the code that aggregates suggestions and the existing suggestion shape; done means each returned suggestion includes a stable classification such as the requested type values, with coverage for the existing checks.

Written by the indexing model from the issue text.