The issubclass should always be false in Python3, so this change
should not change behavior.

Before:
```
In [1]: np.dtype({"names": ["a"], "formats": [int], "offsets": [2]})
Out[1]: dtype({'names':['a'], 'formats':['<i8'], 'offsets':[2], 'itemsize':10})
```

After:
```
In [1]: np.dtype({"names": ["a"], "formats": [int], "offsets": [2]})
Out[1]: dtype({'names': ['a'], 'formats': ['<i8'], 'offsets': [2], 'itemsize': 10})
```

* Allow switching back to old dtype printing format.

* Add changelog.

This adds a C-side scaled float (actually double), that is available
as:

     SF = np.core._multiarray_umath._get_sfloat_dtype()

It supports different scaling factors:

    a = np.arange(10.).astype(SF(2.))
    b = np.arange(10.).astype(SF(0.5))

and casting from double (used there).

This commit makes tiny changes in NumPy to support correct printing.

BUG, ENH: fix array2string rounding bug by adding min_digits option

Co-authored-by: Sebastian Berg <sebastian@sipsolutions.net>

Co-authored-by: Sebastian Berg <sebastian@sipsolutions.net>

Fixes #18609