[8pt] Update CatFIM to accommodate sites where the flood levels are elevations, not stages

[EmilyDeardorff]

Based on vlab ticket #141569.

At site prdk2 in Paradise, KY, the observations, forecasts, and flood thresholds are elevations rather than stage values. This means that stage-based CatFIM is inundating that site up to 378 ft for 'Action' stage, which is creating an inundated area over 10 miles wide.

Write a workaround into the stage-based CatFIM code to detect and mitigate this issue. Then test that this workaround is only mitigating sites with this issue (and not areas where the zero datum is just a very small number).

The workaround should look something like this:

if action_stage > zero_datum:
action_stage =- zero_datum

The following sites will likely also have this issue: CENI3, CLFI3, PNXP1, PVLK2, and ISVO1.

Current behavior

Expected behavior

A much smaller inundated area is expected for action stage at this site.