Divide bequest pools by actual group populations, not birth shares

[marcelolafleur]

Fixes #1186.

With demographics that vary across lifetime income groups (#1165), each group's population share no longer equals its birth share lambdas[j] and survivorship differs by group. The use_zeta = False branch of household.get_bq still divided each group's bequest pool by lambdas[j], so bequests received did not sum to bequests left. In an OG-ZAF calibration with South African mortality gradients, households collectively received about 6.5% more bequests than existed, and the steady state failed the aggregate resource constraint by about 1% of GDP.

The fix: divide each group's pool by its actual population share: omega_SS[:, j].sum() in the steady state, per-period omega sums along the time path, which is what the use_zeta = True branch already does. When demographics are common across groups the two denominators coincide, so results are unchanged in every pre-#1165 use case: all existing get_bq tests pass without modification.

Tests: adds a conservation test with a non-separable omega (group population shares deliberately different from lambdas) checking that receipts aggregate back to the pool for SS and TPI, by-group and pooled. tests/test_household.py passes 52/52 locally.

cc: @rickecon @jdebacker

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 72.68%. Comparing base (b9b33ac) to head (621b97b).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1187      +/-   ##
==========================================
+ Coverage   72.66%   72.68%   +0.01%     
==========================================
  Files          22       22              
  Lines        5726     5730       +4     
==========================================
+ Hits         4161     4165       +4     
  Misses       1565     1565              

Flag	Coverage Δ
unittests	72.68% <100.00%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
ogcore/household.py	90.90% <100.00%> (+0.13%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.