gh-36265: Raise ValueError early in Bijectionist.set_value_restrictions() when restriction is empty

[Chaitanya140904]

Description

Bijectionist.set_value_restrictions() accepted value restrictions that produced an
empty intersection with Z without complaint. The ValueError was only raised much
later — deep inside _compute_possible_block_values() when solutions_iterator() was
already running — making it very hard to trace back to the original typo.

What problem does this solve?

# Before this fix — error fires only when iterating solutions
sage: A = [1, 2, 3]
sage: b = Bijectionist(A, A)
sage: b.set_value_restrictions((1, [4]))   # silently accepted
sage: next(b.solutions_iterator())         # error raised here, far from the typo
...
ValueError: no possible values found for singleton block [1]

A second common pitfall: passing a tuple of values instead of a list
(iterables as elements make this especially confusing):

sage: A = [(1, 2), (3, 4)]
sage: b = Bijectionist(A, A)
sage: b.set_value_restrictions(((1, 2), (3, 4)))  # (3, 4) treated as iterable of values

### How it is fixed

`set_value_restrictions()` now checks the intersection immediately and raises a
descriptive `ValueError` right at the point of the mistake:

```python
# After this fix
sage: A = [1, 2, 3]
sage: b = Bijectionist(A, A)
sage: b.set_value_restrictions((1, [4]))
Traceback (most recent call last):
...
ValueError: the value restriction for element 1 is empty:
none of the given values [4] lie in Z = {1, 2, 3}

[mantepse]

I like your change to bijectionist.py, but there are two files that do not belong to the commit.