`records` rule will sometimes cause records to start violating `columns` parameter

With the following config:

```
steps:
- records:
equals: "same_line"
first_field: "indent 4"
field_comment: 2
deriving: 4
via: "same_line"
break_enums: true
break_single_constructors: false
curried_context: false
columns: 100
```

It converted our existing (desired) code:

```
data Api r = Api
{ _listDocuments :: r :- "list-documents"
:> QueryParam' '[Required] "id" ThreadId
:> Get '[JSON] [Document0]
, _uploadDocument :: r :- "upload-document"
:> QueryParam' '[Required] "id" DocumentId
:> ReqBody '[JSON] MisoString
:> Post '[JSON] ()
} deriving (Generic)
```

To:

```
data Api r = Api
{ _listDocuments :: r :- "list-documents" :> QueryParam' '[Required] "id" ThreadId :> Get '[JSON] [Document0]
, _uploadDocument :: r :- "upload-document" :> QueryParam' '[Required] "id" DocumentId :> ReqBody '[JSON] MisoString :> Post '[JSON] ()
}
deriving (Generic)
```

It would be nice if there was an option to leave the former alone.

Start by tracing the records rule and the columns-width handling using the configuration and before-and-after examples in this issue. Determine how the rule can preserve the existing multiline layout without violating the columns setting, then add coverage for this configuration and verify the formatted output remains within the requested behavior.

Written by the indexing model from the issue text.