[napattan]

Changes proposed in this PR:

• Raise a descriptive ValueError in world_bank() (climada/util/finance.py) when download_world_bank_indicator() returns an all-NaN series (close_val.empty after .dropna()), and include ValueError in income_group()'s fallback exception tuple so missing World Bank income group rows fall back cleanly to nat_earth_adm0().
• Extend nat_earth_adm0() in climada/util/finance.py to check fallback ISO code attributes ("ADM0_A3", "ISO_A3", "WB_A3", "ADM0_A3_US") (resolving ESH and ALA lookup failures in missing worldbank indicators #1331) and distinguish missing country records (close_val is None) from legitimate 0.0 GDP records (such as HMD), logging a warning when Natural Earth reports 0.
• Guard world_bank_wealth_account() in climada/util/finance.py against all-NaN NW.PCA.TO indicator series and np.isnan(fac) when wealth2gdp() returns (None, np.nan), falling back to fac = 1.0 with a warning instead of propagating NaN.
• Add unit tests in climada/util/test/test_finance.py (test_ne_fallback_iso_and_zero_gdp_pass, test_wb_empty_indicator_raises_value_error, test_pca_missing_wealth2gdp_factor_pass).

Reproducible Verification Script (#1331)

Below is a self-contained script verifying all four failure modes reported in #1331 (world_bank empty indicator ValueError, income_group fallback to Natural Earth, nat_earth_adm0 fallback ISO attributes & zero-GDP handling, and world_bank_wealth_account NaN wealth-to-GDP factor fallback):

from unittest.mock import patch
import numpy as np
import pandas as pd
from climada.util.finance import (
    _nat_earth_shp,
    income_group,
    nat_earth_adm0,
    world_bank,
    world_bank_wealth_account,
)

shp_file = _nat_earth_shp("10m", "cultural", "admin_0_countries")

# 1. Fallback ISO attributes (ESH under WB_A3/ADM0_A3_US, ALA under ISO_A3) & 0.0 GDP (HMD)
yr_esh, val_esh = nat_earth_adm0("ESH", "GDP_MD", "GDP_YEAR", shp_file=shp_file)
yr_ala, val_ala = nat_earth_adm0("ALA", "INCOME_GRP", shp_file=shp_file)
yr_hmd, val_hmd = nat_earth_adm0("HMD", "GDP_MD", "GDP_YEAR", shp_file=shp_file)
print(f"1. nat_earth_adm0: ESH=({yr_esh}, {val_esh}), ALA=({yr_ala}, {val_ala}), HMD=({yr_hmd}, {val_hmd})")
assert (yr_esh, val_esh) == (2007, 9.07e08)
assert (yr_ala, val_ala) == (0, 4)
assert (yr_hmd, val_hmd) == (2016, 0.0)

# 2. All-NaN World Bank indicator raises ValueError (and income_group falls back to Natural Earth)
empty_series = pd.Series([np.nan, np.nan], index=[2019, 2020])
with patch("climada.util.finance.download_world_bank_indicator", return_value=empty_series):
    try:
        world_bank("GIB", 2020, "NY.GDP.MKTP.CD")
    except ValueError as err:
        print(f"2. world_bank all-NaN raised expected ValueError: {err}")
    ig_yr, ig_val = income_group("ALA", 2020, shp_file=shp_file)
    print(f"3. income_group('ALA', 2020) fallback -> year={ig_yr}, group={ig_val}")
    assert ig_val == 4

# 3. Missing wealth2gdp factor (NaN) falls back cleanly without returning NaN
with patch("climada.util.finance.download_world_bank_indicator", return_value=empty_series), \
     patch("climada.util.finance.wealth2gdp", return_value=(None, np.nan)), \
     patch("climada.util.finance.gdp", return_value=(2015, 5.0e10)):
    wb_yr, wb_val, q = world_bank_wealth_account("LBY", 2015, no_land=True)
    print(f"4. world_bank_wealth_account('LBY', 2015) fallback -> year={wb_yr}, val={wb_val}, q={q}")
    assert q == 0 and not np.isnan(wb_val)

This PR fixes #1331

PR Author Checklist

• Read the Contribution Guide and AI Policy
• Correct target branch selected (if unsure, select develop)
• Descriptive pull request title added
• Source branch up-to-date with target branch
• Documentation updated
• Tests updated
• Tests passing
• No new linter issues
• Changelog updated

PR Reviewer Checklist

• Read the Contribution Guide and AI Policy
• CLIMADA Reviewer Checklist passed
• Tests passing
• No new linter issues