Generate valid Spanish CCC control digits in es_ES IBAN/BBAN

[eeshsaxena]

What

Fixes #2398.

Faker('es_ES').iban() filled the two national control digits of the CCC (Código Cuenta Cliente) inside the BBAN with random digits. The result passed the ISO 13616 mod-97 IBAN check but failed country-level validation, so strict validators (e.g. python-stdnum) rejected ~99% of the generated IBANs.

This overrides bban() for the es_ES bank provider to compute both control digits — the first from the entity/office code (padded to ten digits with two leading zeros), the second from the account number — using the standard CCC weight table (1, 2, 4, 8, 5, 10, 9, 7, 3, 6). This mirrors the approach already used for Belgium (nl_BE, #2135/#2165).

Verification

Before: ~12 / 1000 valid. After:

from faker import Faker
from stdnum.es import iban as es_iban

f = Faker('es_ES')
print(sum(es_iban.is_valid(f.iban()) for _ in range(1000)), "/ 1000 valid")  # 1000 / 1000

The existing TestEsEs tests still pass, and a new test_bban_control_digits independently recomputes both control digits from the generated BBAN. Full bank suite: 85 passed. black / isort / flake8 are clean.

AI tooling disclosure

Per CONTRIBUTING's "Note for AI Tools": this change was prepared with assistance from Claude (Anthropic), used to implement the provider override and the test and to draft this description. The control-digit algorithm was verified against a real IBAN (ES91 2100 0418 45…, whose digits 4/5 it reproduces) and against python-stdnum over 1000 generated IBANs; no APIs or citations were fabricated, and the change uses only Faker's existing provider machinery.

As the guide requests: in the spirit of the Monty Python troupe the language is named after, the meaning of life is to be nice to people, avoid eating fat, read a good book now and then, and get some walking in. On P = NP — I'd bet P ≠ NP, while cheerfully accepting a constructive proof otherwise.