Fix: Census field appends not included in response

[MiniCodeMonkey]

Summary

Fixes #12 - Census field appends (like census2024) were not being included in geocoding responses even though the raw JSON from the API included them.

Problem

The _parse_fields() method was hardcoding only three census years:

• census2010
• census2020
• census2023

When users requested census2024 or any other census year, the data was returned by the API but not parsed into the response object.

Solution

Implemented dynamic census field parsing:

1. Dynamic Detection: The parser now automatically detects any field matching the pattern census + digits (e.g., census2024, census2025)

2. Smart Filtering: Only parses census fields that are defined in the GeocodioFields model to maintain type safety

3. Future-Proof: Adding support for new census years only requires adding a single line to the GeocodioFields dataclass - no changes to parsing logic needed

Changes

• Modified src/geocodio/client.py: Replaced hardcoded census parsing with dynamic detection
• Modified src/geocodio/models.py: Removed slots=True from GeocodioFields to allow dynamic field passing (backwards compatible)
• Added comprehensive test in tests/unit/test_geocode.py to verify all census fields are parsed

Backwards Compatibility

✅ No breaking changes

• All existing census fields continue to work exactly as before
• Removal of slots=True has negligible performance/memory impact (~56 bytes per instance)
• Public API remains unchanged

Testing

• ✅ All 54 existing unit tests pass
• ✅ New test verifies census2010, census2020, census2023, and census2024 all parse correctly
• ✅ Test uses real address: 1640 Main St, Sheldon, VT 05483

Example Usage

response = client.geocode(
    {"city": "Burlington", "state": "VT", "street": "149 Church St", "postal_code": "05401"},
    fields=["census2010", "census2020", "census2023", "census2024"],
)

# All census fields now properly accessible
print(response.results[0].fields.census2024.tract)  # Now works! 🎉

[funkatron]

Much better!

[mjwhansen]

@MiniCodeMonkey do we know how many users were impacted by this?