Update stochastic physics generation interval

[MatthewPyle-NOAA]

In a retro REFS run (mem 3 , 20240518/18Z), the initial run with lndpint=180 failed about 2.5 hours into the forecast. A rerun with lndpint=72 ran to completion? Properly update so all ensemble namelists will be generated with a smaller than 180 s value for lndpint.

[ShunLiu-NOAA]

Will this potentially change the solution or use more computational resource?

[MatthewPyle-NOAA]

Change the solution slightly, and will add more time to the run.

[MatthewPyle-NOAA]

After discussion with Shun, may want to have this shortening as a backup option, only used if things fail. But might want to set the default to an intermediate value, maybe lndpint=108.