Keep string data as python str

[jgrewe]

With b3c21c3 we fixed the reading of 2D string data. Unfortunately, this implicitly converted python strings to np.str_. With this pr, we keep them as python strings.

fixes #534

[JuliaSprenger]

Would it make sense to add a test for this fix?

[jgrewe]

@JuliaSprenger thanks. Added a test, if you're happy, I guess you may approve and merge ;)

[JuliaSprenger]

I don't have the rights to merge. @mpsonntag can you?

[mpsonntag]

Looks good to me

[JuliaSprenger]

Thanks! Are you planning a minor release with the fix soon? In that case I could adjust the neo dependency already now.

[jgrewe]

actually yes. As soon as the other PRs pass and are merged. I need the changes in PR #537 for some current stuff.

That pr raises an interesting question: I think it fair to raise OutOfBound exceptions in our index_of methods but the methods that should return a chunk of data such as data_array.get_slice should not. Imho, they should simply return an invalid (empty) DataView, if there is actually no data in that slice. PR #537 goes that way for event data where it can easily happen that there are simply no events in the segment... For sampled data we still throw the exceptions...
If you have an opinion on that, it would be nice to know ;)

[mpsonntag]

Rebase LGTM

[JuliaSprenger]

Hi again! It seems all other PRs are merged. So is there a version 1.5.3. release coming soon?

[jgrewe]

@JuliaSprenger Done! :)

[JuliaSprenger]

Thank you :)