so the handler is not called while looping through the
list of these replies to abort them. The handler must
not be called in that situation because it mutates the
list.
* Ensure previous long polling requests for events are aborted
before connecting again
* Ensure results from previous requests are always discarded
after aborting to reconnect
* Rely not only on the FolderErrors event
* Request pull errors when opening the dialog
* Use "pullErrors" from FolderSummary
* Move code for directory error dialog into
its own class
This should fix that sometimes obsolete errors
were still displayed or actually present errors
missing.
* Discard folder errors older than the last "sync" state
* Unfortunately we don't know the the time of the last
sync from the beginning. To prevent showing obsolete
errors in this state, discard all errors if the status
changes to something other than "out-of-syn".
