fix missing datasource error message

[timifasubaa]

When a datasource is deleted, slices show a cryptic error message in image 1.
This PR fixes it so it shows a more intelligible error message as in image 2.

@john-bodley @michellethomas

[codecov-io]

Codecov Report

Merging #5019 into master will decrease coverage by <.01%.
The diff coverage is 66.66%.

@@            Coverage Diff             @@
##           master    #5019      +/-   ##
==========================================
- Coverage   77.34%   77.34%   -0.01%     
==========================================
  Files          44       44              
  Lines        8665     8668       +3     
==========================================
+ Hits         6702     6704       +2     
- Misses       1963     1964       +1

Impacted Files	Coverage Δ
superset/views/core.py	74.64% <66.66%> (-0.02%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5a64b3f...63115fb. Read the comment docs.

[john-bodley]

Should we cascade deleted datasource and delete the corresponding slice(s)?

[mistercrunch]

I don't understand how datasource_id == 'None' comes into being in the first place. Can we catch the issue upstream?

[timifasubaa]

True. I was also confused at the beginning. datasource evaluates to 'None__Table', so splitting that gets 'None'. I'll investigate upstream.

[timifasubaa]

@john-bodley In the case where I found the issue, the user wasn't aware the underlying datasource was missing. I think deleting would have been even more confusing.

[timifasubaa]

@john-bodley @mistercrunch Upon furthher investigation, I think the current approach is best for the following reaosns

1. datasource_info is the first place the datasource is checked from the form data. And it's the first place

2. The method intentionally does not fail because the error is handled downstream (https://github.com/apache/incubator-superset/blob/master/superset/viz.py#L64), where the intelligible error message (Viz is missing a datasource) is displayed.

[mistercrunch]

Can we change the error message to something like "The data source associated with this chart no longer exists"?

Also, should we raise right then (asap) instead of waiting for raising in viz.py

[mistercrunch]

About cascading, I think that would be the desired behavior as long as there's clarity on what is happening. There should be a confirmation screen with all the associated objects that are going to be deleted, and the person performing the operation should have write/delete access to all these objects.

[timifasubaa]

Agree with raising early. Done that.

I agree with the cascading idea of forcing users to delete all dependent slices before deleting the datasource but I'll push it to a future PR.