Guess how long a Sparv process will take to run

[anne17]

Many test users have expressed the desire to get an estimate of how long they will have to wait until their corpus is done processing. Can we guess this somehow? How long it takes depends on:

• place in the queue
• total corpus size (amount of files, file sizes, total size)
• which annotators are run (stanza, swener and compound analysis take the longest)
• how clean the data is (a lot of ORC-trash often increases processing time)
• if there are many long strings in the texts it will take longer time to process
• ...?

If we have a time estimate per corpus we should be able to get an estimate for a queued corpus.