Can we try to simplify the calculation function when interpolation is not required？

[ShareRingNS]

interpolate_1d often requires iteration or other complex forms, and brings many bugs. Why not remove this function when I have the required pressure layers and corresponding other physical quantities? I refactored the function and removed this part when using it. Maybe the interface can be set in the official source code to simplify the calculation.

[dopplershift]

I'd be supportive of a Pull Request that refactors some calculations to support this. In my mind, what we want is a function, say maybe called extract_levels() that, given the pressure levels and the source arrays:

1. Check if the desired levels are in the data, and if so directly pull them out
2. Otherwise, fall back to interpolate_1d to calculate the data

I don't think this is important from a bug standpoint, as I wouldn't say interpolate_1d "brings many bugs". There's one bug right now, and unless it's particularly problematic, it should be fixed in our next release, which is coming out imminently.

The benefit of the refactor is it is much more performant (avoids significant calculation to do the interpolation), and in the cases where we're using it in sounding indices, the interpolation is unnecessary since they're mandatory levels that 99% of data I've encountered has.

[ShareRingNS]

Yeah, I think it is more friendly for beginners, because I spent a lot of time to distinguish which calculations can be directly performed matrix operations without interpolation