OAI endpoint error handling

[antmoth]

We should kindly inform the user that there is a problem if an attempted OAI harvest finds:
(1) a down endpoint
(2) no records

[antmoth]

^ no records scenario maybe taken care of

handling "endpoint down or unreachable" might require a completely different code intervention. I knew there was a reason I thought those should be separate issues :D

[antmoth]

I'm not entirely sure that marking the job as invalid when it just has no records is the right thing to do... Maybe I should add a 'warning' state, 'errored' or something.

[antmoth]

The problem with 'down endpoint' is that when I try to, e.g., put in an invalid endpoint URL, the job seems to never stop trying, and to run forever? @ghukill

[antmoth]

Could run a health test on the endpoint/ping it before the Spark task runs. Error language needs to be friendly