(invalid) addresses with > breaks parser

I got an email with an invalid `To` field.
Notice the `>` at the end of the address.
(Example is stripped down to the relevant parts):

```Date: Fri, 12 May 2017 17:49:04 +0200
From: Example
To: info@example.com>
Subject: =?UTF-8?Q?Test_Mail?=

Test Mail=0D
=0D
```
Getting:
```ruby
Mail.new(..).to # => "info@example.com>"
```
expected:
```ruby
Mail.new(..).to # => ["info@example.com>"]
```
Thunderbird automatically strips the invalid `>` so this would be a bonus.

One more example with multiple addresses:
```
Date: Fri, 12 May 2017 17:49:04 +0200
From: Example
To: info@example.com>, info2@example.com
Subject: =?UTF-8?Q?Test_Mail?=

Test Mail=0D
=0D
```
Getting
```ruby
Mail.new(..).to # => "info@example.com>, info2@example.com"
```
expected:
```ruby
Mail.new(..).to # => ["info@example.com>", "info2@example.com"]
```

As those emails really exist (the sender used the 'goneo Webmail', a german mail-hosting company) a fix within the gem which helps other ppl, too, would be much appreciated.

Please let me know if this would be a possible option.

Reproduce the issue with the two raw To-field examples and inspect the address-parsing entry point reached by Mail.new(..). Compare the current scalar results with the expected address arrays, including the single trailing > and the comma-separated case. Done means both examples parse into the expected recipient lists without breaking valid addresses.

Written by the indexing model from the issue text.