Fix url with slices

[vallsv]

This was just not implemented.

Changelog:

[t20100]

Indeed support of slice was missing.

Let's keep helper functions (slice_to_string and slice_sequence_to_string) private to make it clear it's not part of the API.

[vallsv]

Here is with fixes.

slice_sequence_to_string is public because i need it at another place.

[t20100]

One corner case: How does it work with () as slicing?

[vallsv]

(basically there is the same bug in UrlSelectionTable, cause it was a copy paste, i prefer to use the same base code for now)

[vallsv]

@t20100 what () should do? i have no idea.

[vallsv]

Actually

• () -> slice=
• (1, ) -> slice=1

I suggest to rework that to have

• () -> slice=,
• (1, ) -> slice=1,

but i am really not sure, it's maybe better to change that in another PR

[t20100]

what () should do?

From https://numpy.org/doc/stable/user/basics.indexing.html :

An empty (tuple) index is a full scalar index into a zero-dimensional array. x[()] returns a scalar if x is zero-dimensional and a view otherwise. On the other hand, x[...] always returns a view.

Basically it returns everything but it also works with scalars:

In [6]: s = numpy.array(1)
In [7]: s[()]
Out[7]: 1
In [8]: s[...]
Out[8]: array(1)
In [9]: s[:]
---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)

In [10]: a = numpy.array([1])
In [11]: a[:]
Out[11]: array([1])
In [12]: a[...]
Out[12]: array([1])
In [13]: a[()]
Out[13]: array([1])

[t20100]

Acutally it should be OK as it is:
() corresponds to no slicing.

[vallsv]

Ok, nice, i prefer not to touch this semantic.

It's just that there is an inconsistency because 1 is the same as (1,), which is a bit weird. But with this 2 entries the result should also be the same with numpy slice, so i guess it's fine.