[wip] rustify parts of job and imap handling, introduce Delay and ImapResult enums like in C

[hpk42]

removes "ok_to_continue" logic -- for reviewing lookings at non-cargo-fmt commits might be better. Or reviewing the end result of the changed functions and comparing them with C.

[r10s]

maybe we should do a

return if !self.should_reconnect() {
                ImapResult::Failed
            } else {
                ImapResult::RetryLater
            };

here as well. otherwise, if network does away during copy, the message is never tried to copy again.

maybe it is also better to add the same check after mv-failing already - if it turnes out, network is away after move, trying over with copy+delete mayl raise the risk in a flaky network situation that one of these part fails.

[hpk42]

closing this PR but the branch remains in active development -- working with @r10s on larger scale changes on it: introducing perform_mvbox_jobs to make all jobs that do things with mvbox actually use the mvbox thread -- this way inbox thread does not have to select mvbox folder and then select back etc. -- it not only decreases network traffic but also decreases the dependency on interrupt_idle working in several situations. (right now, all imap jobs on all folders are running the perform_imap_jobs (to be renamed to perform_inbox_jobs at some point) action). This all came about because we couldn't do certain things because of hangings in the tests because we can't reliably interrupt idle, see jonhoo/rust-imap#132