Location code regex are not correct

[marmg]

Hello, I see in validation.py that the code for checking if it's a subnational code is:

def is_subnational2(value):
    return re.match(r"^[A-Z]{2}-[A-Z0-9]{2,3}-[A-Z0-9]{2,3}$", value)

However, some location codes (like in Ireland) contain only 1 letter: IE-L-DN

[pkesling]

I've encountered this issue as well.

I did some quick verification across all regions worldwide (down to the subnational2 level) comparing the current validation versus the proposed ("fixed") validation. The proposed fix will solve the problem for all current country, subnational1, and subnational2 regions.

# ./.venv/bin/python analyze_region_code_validation.py ./worldwide_region_codes.json
CURRENT VALIDATION RULES (country=r'[A-Z]{2}', subnational1=r'[A-Z0-9]{2,3}', subnational2=r'[A-Z0-9]{2,3}')
Results of regional code validation (regex) assessment:
 >> Country codes: total=253, valid=253, invalid=0
 >> Subnational1 codes: total=3698, valid=3648, invalid=50**
 >> Subnational2 codes: total=16693, valid=16693, invalid=0

PROPOSED VALIDATION RULES (country=r'[A-Z]{2}', subnational1=r'[A-Z0-9]{1,3}', subnational2=r'[A-Z0-9]{2,3}')
Results of regional code validation (regex) assessment:
 >> Country codes: total=253, valid=253, invalid=0
 >> Subnational1 codes: total=3698, valid=3698, invalid=0
 >> Subnational2 codes: total=16693, valid=16693, invalid=0

Would love to see the pull request (#28) approved.