Fix versionadded tags for v3.4.3

re.split("\s*") has changed from splitting on 1 or more spaces in Python 3.6
(with a deprecation warning) to splitting on 0 or more spaces in Python 3.7.

In Python 3.7 this means it splits at every character.

With Python 3.6:

python -c "import re; print(re.split('\s*', '#define H5_VERS_MAJOR    1'))"
/opt/conda/lib/python3.6/re.py:212: FutureWarning: split() requires a non-empty pattern match.
  return _compile(pattern, flags).split(string, maxsplit)
['#define', 'H5_VERS_MAJOR', '1']

With Python 3.7b3:

python -c "import re; print(re.split('\s*', '#define H5_VERS_MAJOR    1'))"
['', '#', 'd', 'e', 'f', 'i', 'n', 'e', '', 'H', '5', '_', 'V', 'E', 'R', 'S', '_', 'M', 'A', 'J', 'O', 'R', '', '1', '']

Python 3.7 fix for re.split() in version detection

Remove unused import and fix ImportError with numexpr 2.6.5.dev0

While fixing fall-outs of changes to `numpy.rec.array`
(see numpy/numpy#10344) I evidently missed
a spot which is now causing trouble. See

#664

I think I noted this in the aforementioned PR in a comment.

This PR is to make sure that `initRecArray` method produces list of
tuples, rather than list of lists, as an argument to `numpy.rec.array` constructor.

MAINT: Fixed test failures with Python 2.7 and NumPy 1.14.3

Include citation in repo

Updating from original project

Changes author to PyTables Developers Team