text garbled for mail subject decoding

https://github.com/mikel/mail/blob/8fbb17d4d5364c77cc870769d451bc2739b3a8ce/lib/mail/encodings.rb#L123-L138

here is the original data of subject
```
Subject: =?iso-2022-jp?Q?=1B=24B0F7o=3EpJs=1B=28B=5B=2D01_=1B=24?=
=?iso-2022-jp?Q?BBg=3Cj=3Bq=3B=3A1=3FMQ2q=3CR=1B=28B_=2D_=1B=24B=3B?=
=?iso-2022-jp?Q?q=3B=3A1=3FMQ=24K=24=2B=24=2B=24k=1B=28BDWH=1B=24B=24?=
=?iso-2022-jp?Q?NFs=3C=213=2BH=2F6HL3=1B=28B=5D?=
```

when i use `mail.subject`,it returns
`"案件情報[-01 �BBg

Start with lib/mail/encodings.rb#L123-L138 and inspect how Encodings.value_decode handles folded Subject lines. Reproduce the issue with the provided ISO-2022-JP encoded subject, comparing the folded and manually concatenated forms. Done means mail.subject returns the complete Japanese subject for the original folded header.

Written by the indexing model from the issue text.