DataSet.get_data bugs with start/end arguments

[astafan8]

Fixes #1384.

There are two bugs in the underlying get_data function of sqlite_base.py:

• 0 are treated as None
• sql query has a typo in the where clause for start for the case where both start and end are not None

This PR fixes both.

Note that the current behavior suggests a 1-based indexing (unlike python 0-based), with both start and end indexes being included into retuned data. This behavior is captured by the test suite.

Ideally, the very same test should be done for get_data from sqlite_base.py but did not make it to this PR in order to avoid code duplication.

Note that get_data from sqlite_base.py was also refactored to reduce the possibility for bugs like the ones mentioned above. In case it made the function less readable, the refactoring can be reverted.

[codecov]

Codecov Report

Merging #1386 into master will increase coverage by 0.06%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1386      +/-   ##
==========================================
+ Coverage   73.38%   73.45%   +0.06%     
==========================================
  Files          79       79              
  Lines        9282     9283       +1     
==========================================
+ Hits         6812     6819       +7     
+ Misses       2470     2464       -6

[astafan8]

@QCoDeS/core ready for review.

[WilliamHPNielsen]

I think this looks good. I have the feeling that the test could be written without the mechanisms that almost (but not quite, which is important here) duplicate the existing fixtures, but after discussions with @astafan8, I feel that such a rewrite would probably be too much work for too little actual gain.

The non-test part looks like a clear improvement.

[jenshnielsen]

Looks good, left a few small comments