changed mailimap's member `complex_command_tag_enabled` to `is_163_workaround_enabled`.

changed 
` int mailimap_is__complex_command_tag_enabled(mailimap * imap); `
` void mailimap_set_complex_command_tag_enabled(mailimap * imap, int enabled);`
to
`int mailimap_is_163_workaround_enabled(mailimap * imap);`
`void mailimap_set_163_workaround_enabled(mailimap * imap, int enabled);`

changed 
` int mailimap_is__complex_command_tag_enabled(mailimap * imap); `
` void mailimap_set_complex_command_tag_enabled(mailimap * imap, int enabled);`
to
`int mailimap_is_163_workaround_enabled(mailimap * imap);`
`void mailimap_set_163_workaround_enabled(mailimap * imap, int enabled);`

163 mail server will reject the command which with a number tag.

Fix IMAP ID send.

Use mailimap_quoted_send rather than mailimap_astring_send when
sending names and values inside mailimap_id_param_send.
This means skipping the "workaround for buggy Courier-IMAP that
does not accept quoted-strings for fields name but prefer atoms"
and instead always sending quoted strings.

RFC 2971 (with reference to RFC 2060) is clear that these fields
may only be quoted strings or literals, not atoms.  Yahoo's server
rejects this command when atoms are sent (and it is correct to do so).

We shouldn't break compat with well-behaving servers as a workaround
for buggy ones, so using quoted strings here is the best choice.

Added MAILSMTP_ERROR_AUTH_TIMEOUT

kCFStreamPropertySSLPeerTrust fail on OS X<=10.8 - infinite loop

typo fix…

removing fprintf

on OS X <= 10.8 (maybe 10.9 too), the kCFStreamPropertySSLPeerTrust
method works the first time, but always returns NULL after, causing an
infinite loop. This change falls back to the old
kCFStreamPropertySSLPeerCertificates method if the first one returns
NULL.

This reverts commit 975eebbe04.