Bugfix/nbi err handling

Fixing the error handling as suggested in #45.

I did not change it in all the places, and I won't do it now

requested review from @ffelici

approved this merge request

Are you sure that empty dim is in this case always the same as mds error?

but... I modified it as you asked. I removed the check on the dim

For the other cases you also check also the data. But if you get an MDS error for this node in cases NBI was not requested, this is ok.

I am still surprised that the best way to check whether NBI was used is to try to ask for data from node and see if you get an error. Don't we have a flag written somewhere (that does not return an error when looking it up) telling us the NBI was off?

Why are you looking at lines that have been removed?

I'm not looking at the line that has been removed I'm commenting on the newly added statement.

Note that this code is being excluded from the tests. Can we add a case to the tests that result in these lines being called?

I am not sure what you refer to. Why are you looking at lines that have been removed? Those lines in principle are always there

I don't know about the tests. And I think that is another issue to open, not related to this MR

If you look in the 'changes' tab of the MR, you see which lines of code are being covered by the tests and which are not. We should try to have as much as possible of our code covered by tests.

The vertical red lines indicate no test coverage

Thank you @vallar I quickly reviewed it

changed the description