Incorporate missing MODIS SCF (snow cover fraction) in Estimate_v61

[bidhya]

In the new version Estimate_v61.jl, we do not handle missing MODIS SCF data.
While this will certainly occur in high latitudes due to polar nights, I observed missing SCF in other places as well.

[bidhya]

Prior Approach

set MSCF[i] = 1 for missing data, then subsequently, set
σWRFG[i] = 1e9 (a really high value).
Jack made updates to this approach and is handled a bit differently (Estimate_v60.jl)
The problem in incorporating this strategy for the new version (Estimate_v61.jl) is that WRFG (Gprior or heat flux) is not used in new version.

[bidhya]

skipmissing julia function may necessary but not sufficient. For example, replace

median(SCF_inst[i-twindow:i+twindow]) with 
median(skipmissing(SCF_inst[i-twindow:i+twindow]))

to handle missing (Nans) data