E1131 TermTagger DOWN, when termtaggers are only temporarily not reachable

If all termTaggers are set to "temporarily not reachable", the translate5 import throws an error.

This should not happen immediately, but only after a configurable time (Zf_configuration).

This avoids imports to crash, if all termTaggers are not reachable in reasonable time, because they have to much work to do.

Implementation Notes:

Idea 1: make a sleeped loop on getting the above error: Not possible since the error is in the queue call, there fore multiple processes in the above sleep will be possible and will block apache processes.

Idea 2: The concrete problem should be better investigated. Therefore a worker dump should be created when the next error occurs. Either manually if we get an error, or we have to implement a logging mechanism there. Reason: Through slots and maxParallelWorkers = 1 it should not happen that multiple tag processes should use the same worker, they should be worked sequentially. So if the worker log shows if this is right, or if there is another deeper problem. And this shows that the timeout is causing the DOWN is probably coming because one term tagging request is just running to long. The timeout is currently 180seconds, it is increased to 300 seconds on the affected instance. Either this solves the problem, or we have to investigate the worker table content. After that we can decide how the best solution will look like.

Hint: Currently a DOWN is a timeout and if the server is not reachable at all. Probably this must be split up in separate exceptions for different handling.