End-to-end BGE runs 1–2 config and mount preflight

[alxsmith]

Overview

• Wire BGE end-to-end run config for MD: electric tariff maps, null gas tariff, scenario capacity-MC paths, and default fixed-charge cleanup.
• Fail fast when CAIRO would write under /data.sb while the s3fs mount is down (Justfile preflight + assert_output_dir_is_mounted in run_scenario.py, with tests).

Reviewer focus

• Mount preflight: Justfile check vs Python path-aware check — intentional defense-in-depth so direct run_scenario.py invocations are still guarded.
• BGE tariff maps and null gas tariff: confirm they match the intended run-1/run-2 assignment for md_20260732a_r1-style batches.

Non-obvious details

• Delivery-only BGE scenario rows previously pointed path_supply_capacity_mc at NY zero parquet paths; corrected to MD.
• CAIRO cannot write to s3:// directly (pathlib/open()), so outputs depend on /data.sb being mounted. Unmounted mounts previously caused silent local-disk writes.

Closes #490

Made with Cursor

[alexhyunminlee]

One small thing. For the precalc runs, could we change *_pc suffix at the end to just *_precalc? I was confused at first that this meant percustomer, and also just spelling out precalc matches the *_default naming patterns.

[alexhyunminlee]

Also, just so I'm on the same page, we don't have the hp_vs_nhp revenue requirement files and the tariff's yet, right? (because this PR is jsut for runs 1+2) Or are they missing from this PR?

[alexhyunminlee]

Just flagging something that caused me some headache from the NY/RI days: here, the name of the file goes *_calibrated_supply.csv, but the content of the files here (which represent the name of the tariff jsons) goes *_supply_calibrated. Same goes for the other calibrated supply type maps below (e.g. bge_hp_seasonal_calibrated_supply.csv. It might be a good idea to standardize this order throughout so we don't run into unexpected naming errors later on.

[alexhyunminlee]

Looks good to me overall, minor question/comments above.