Bugzilla

Ok, I am able to introduce (simulate) an extreme network delay to my pop3 servers that are not on localhost using tc command
```sudo tc qdisc add dev wlp0s20f3 root netem delay 2000m```
I have one server on localhost which is not affected by this command and remains fast.
I set the poll order like this:
1 for localhost server
2 for server on internet
3 for server on lan
All servers are set up to have a messages ready to be downloaded by TB and all 3 pop3 accounts are "deferred" to Local Folders/Inbox (global inbox). 
I go to Local Folder and select Inbox and click the "get mail" icon. (Note: shift-F5 does the same thing with this setup.)
The messages from localhost server come in immediately and are readable.
After several seconds the messages from server on internet come in and are readable.
After several more seconds the messages from server on lan come in a are readable.

So I conclude from this (and other testing not described) that the ordering has the desired effect and it should fix the comment 0 issue described. 
Hopefully, reporter Pepi can test this too in his "real world" setup.

Ok, I am able to introduce (simulate) an extreme network delay to my pop3 servers that are not on localhost using tc command
```sudo tc qdisc add dev wlp0s20f3 root netem delay 2000ms```
I have one server on localhost which is not affected by this command and remains fast.
I set the poll order like this:
1 for localhost server
2 for server on internet
3 for server on lan
All servers are set up to have a messages ready to be downloaded by TB and all 3 pop3 accounts are "deferred" to Local Folders/Inbox (global inbox). 
I go to Local Folder and select Inbox and click the "get mail" icon. (Note: shift-F5 does the same thing with this setup.)
The messages from localhost server come in immediately and are readable.
After several seconds the messages from server on internet come in and are readable.
After several more seconds the messages from server on lan come in a are readable.

So I conclude from this (and other testing not described) that the ordering has the desired effect and it should fix the comment 0 issue described. 
Hopefully, reporter Pepi can test this too in his "real world" setup.