Improve logging of SSH, TLS issues caused by rekeying

[Castaglia]

The scope of this ticket is to try to improve the logging of failures with encrypted channels (specifically TLS and SSH) that occur due to renegotiations/rekeying. Not all TLS or SSH clients handle such situations well, especially if it is the server, and not the client, that initiates such rekeys.

This came up recently on the ProFTPD users mailing list. The key clue in that case was the "upload failures after 2GB of a 4GB file" when using SFTP; that 2GB size aligned with the default SFTPRekey size parameter. However, this correlation is not immediately obvious, and we want to improve the logging to make it obvious.

[Castaglia]

Note that this is probably related to Bug#4498.

[Castaglia]

Note that as of Bug#4443, TLS renegotiations are disabled by default. That said, having TLS logging that helps to pinpoint TLS issues related to renegotiations (when enabled) would still be a good thing to have.