Use multipart/alternative for email encoding

[gdpelican]

I'm not sure why we're using 7bit currently; maybe ahoy email or premailer or something does this?
https://rails-bestpractices.com/posts/2010/08/05/use-multipart-alternative-as-content_type-of-email/

[gdpelican]

@robguthrie How do you feel about this change? Ref #4022

[robguthrie]

I feel fine about the idea, unsure if it will break stuff in practice.
Give it a go?

[fauno]

Content-Type is different from Content-Transfer-Encoding, and every part can have a different CTE. I just checked the parts and the header and text/plain part are CTE 7bit and the text/html part is quoted-printable... while github notifications are all 7bit.

what happens if i use unicode? can you check this notification? áéíóú, 🙃

[fauno]

by looking at the function that throws the error, i'm guessing it's because of the mixed CTEs [7bit,7bit,quoted-printable]: https://github.com/vstakhov/rspamd/blob/71d52e806a4780d95eb0322544330e31a7a375d9/src/libmime/mime_expressions.c#L1277

i also found a github notification where the main cte was 7bit but being written in correct spanish (with tildes and all) the parts were quoted-printable and it was marked as bad_cte_7bit by rspam, so it may be too picky.

[fauno]

asked on rspamd :)

rspamd/rspamd#1590

[gdpelican]

Okay, we're not going to go with this for now. Hopefully with the move to SES we won't have these deliverability issues going forward.