Fix durations in forcing config files for medium-range lagged ensemble

[mxkpp]

For the medium-range lagged ensemble workflow, this adjusts the realization durations (parameter ForecastInputHorizons) that are written in the forcing configuration files, to match the intended time-windowing.

This also improves error messaging (see below for details).

This should be reviewed in conjunction with related PRs on other repositories:

NGWPC/nwm-rte#52

NGWPC/nwm-msw-mgr#54

NGWPC/nwm-fcst-mgr#18

Changes

• Updated forcing configuration files for medium-range lagged ensemble members to match intended time-windowing of those members (starting with member 2, each runs 6 hours aka 360 minutes longer than its previous member).
• Improved error messaging in err_out_screen to include traceback in to the LOG, in addition to sending the traceback to stderr as had been done before.

Testing

1. Tested initial ensemble setup (not ran to completion) using nwm-rte with command: docker_run python "/ngen-app/bin/bin_mounted/run_forecast.py" -n 2 -dt "2025-09-15 00:00:00" -rname "${fcst_run_name}_mr_le" -fconfig medium_range -le "" ""

Screenshots

Notes

• For each member of the ensemble, the config file parameter RefcstBDateProc is overridden on-the-fly by nwm-fcst-mgr, but ForecastInputHorizons is taken as-is from each forcing config file.

Todos

[jswade-rtx]

Changes look good to adjust the lagged ensemble forcing engine templates.

[mxkpp]

I pushed one more minor improvement to the handling of error messages in err_handler.err_out_screen.