Add fair-default rate design pipeline for NY

[sherryzuo]

Summary

This PR delivers the full fair-default rate design pipeline for NY — the modules, configs, tariff artifacts, tests, and orchestration scaffolding needed to run the three closed-form fair-default strategies (A: fixed_charge_only, B: seasonal_rates_only, C: fixed_plus_seasonal_mc) across all six NY utilities.

Key deliverables:

• utils/mid/compute_fair_default_inputs.py — computes class and HP subclass aggregate inputs (kWh by season, bills, cross-subsidy) from CAIRO baseline outputs
• utils/mid/create_fair_default_tariff.py — closed-form solver for strategies A/B/C via Cramer's rule; returns a modified URDB tariff JSON
• utils/mid/prepare_fair_default_tariffs.py — orchestration layer that runs strategies across all utilities in a batch
• utils/pre/create_fair_default_scenario_yamls.py — generates per-utility fair-default scenario YAMLs (runs 101–124)
• Generalised feasible-line geometry: revenue-sufficiency and cross-subsidy-zero lines in (F, r_sum) space, including revenue-sufficient line for arbitrary seasonal rate combos and cross-subsidy calcs across all seasonal rate combinations
• NY fair-default tariff configs: URDB JSONs, tariff maps, scenario YAMLs, and calibrated tariff JSONs for all six NY utilities (CenHud, ConEd, NIMO, NYSEG, O&R, RGE)
• utils/pre/compute_tou.py — demand-weighted rho_MC input for strategy C
• Extended validate_config.py to require bulk_tx entry in marginal-cost configs
• Context doc: context/methods/tou_and_rates/fair_default_rate_design.md — full math derivation, closed-form strategies, feasibility geometry
• Tests: test_compute_fair_default_inputs.py, test_create_fair_default_tariff.py, test_compute_fair_default_feasible_line.py, test_compute_tou.py
• .gitignore: added .tmp/ for local pytest basetemp
• Removed utils/mid/plot_feasible_line.py (plotting migrated to reports2)

Closes #398

Reviewer focus

• Strategy math: create_fair_default_tariff.py implements a 2×2 Cramer's-rule solve for each strategy — worth a close read to confirm the linear system is assembled correctly relative to the notation in context/methods/tou_and_rates/fair_default_rate_design.md.
• Feasible-line generalisation: 1a57bfc and f7d9c02 extend the feasible-line computation to work across arbitrary seasonal rate combinations, not just the (F, r_sum) plane — verify the parameterisation matches the math doc.
• Config artifacts: the NY utility URDB JSONs, tariff maps, and scenario YAMLs were hand-assembled and then validated by validate_config.py; spot-check NYSEG and RGE since they have the most complex gas/seasonal structures.

Non-obvious implementation details

• Strategy C slaves the winter rate to the summer rate via rho_MC = kWh_win_hp / kWh_sum_hp (demand-weighted, not MC-weighted — see compute_tou.py), which keeps the seasonal differential cost-reflective without requiring a separate MC time-series input at tariff-solve time.
• The prepare_fair_default_tariffs.py orchestration layer is intentionally thin: it calls compute_fair_default_inputs → create_fair_default_tariff per utility per strategy and writes calibrated JSONs back into config/tariffs/electric/, matching the same promotion pattern used by the seasonal-discount workflow.
• Plotting of the feasible line was removed from this repo (commit ad6208c) because it lives in reports2's fair_default_feasible_line.qmd notebook — keeping visualisation out of the platform keeps the dependency boundary clean.