add @cache to freqent model info calls

[JoshCu]

while testing out some bmi code I noticed that certain functions to fetch model info are called by ngen every timestep and sometimes multiple times in a row.
This PR adds @cache to some of these functions

when running gage-10154200 (55 catchments) for 1 year with 6 ensembles:
runtime dropped from 33.619 s ± 0.157 s to 30.941 s ± 0.070 s

when running a single model (ensemble of 1)
14.093 s ± 0.110 s -> 11.436 s ± 0.175 s

[hellkite500]

Python 3.8 has been EoL for over a year now, I wouldn't go out of your way to ensure compatibility.

[JoshCu]

Python 3.8 has been EoL for over a year now, I wouldn't go out of your way to ensure compatibility.

time flies when you're having fun even 3.9 is EoL now, should probably move ngiab off it