Dataset wildcards may lead to "missing data" errors for derived variables

[schlunma]

Describe the bug

When using wildcards in a recipe with derived variables, and not all required input variables for the derivation are available, the tool reports this is an error. Example:

- {project: CMIP6, dataset: "*", exp: historical, ensemble: r1i1p1f1, grid: "*", institute: "*", timerange: 2000/2005, short_name: swcre, derive: true}

will lead to the following error:

ERROR   [2109992] Missing data for preprocessor timeseries/swcre:
  - Missing data for Dataset: rsutcs, Amon, CMIP6, FIO-ESM-2-0, CMIP, historical, r1i1p1f1, gn

Note that the other input variable rsut exists.

I think the tool should not fail here and simply not consider this missing model at all.