FWS handling in message ids

I am curious about the behaviour of the parser for message ids.

```ruby
Mail.read_from_string("Message-ID: < foo\r\n @ example.com\r\n >\r\n").header.fields[0].element.message_ids[0]
```

Yields `" foo @ example.com "`. It means that the library will return different results for message ids which are semantically identical but syntactically different, for example:

- `Message-ID: \r\n`
- `Message-ID: < foo@\t example.com> \r\n`
- `Message-ID: \r\n <\r\n\t foo \r\n @example.com>\r\n`

Unless I am mistaken, all these representations are semantically identical (see RFC 5322 3.6.4., `msg-id = [CFWS] "<" id-left "@" id-right ">" [CFWS]`)and I would expect message ids to be parsed as an object with two fields, `left` and `right`. In all these examples, `left` would be `foo` and `right` would be `example.com`.

Is there a reason behind the current behaviour?

Start by running the Ruby reproduction in the issue and compare its result with the RFC 5322 message-id examples provided. The issue names no files or tests; first locate the message-id parser, then clarify whether the intended outcome is normalization or a structured object with `left` and `right` fields.

Written by the indexing model from the issue text.