i.atcorr: fix out of bound access in trunca, os and iso

[ymdatta]

In trunca(), under 2 if conditionals (if rumu[1] > 0.8 and rmu[1] > 0.94), k and kk values are set to -1 respectively. When these values are used to access an array, it leads to out of bound access, which is undefined behavior in c++.

Similarly in os() and iso(), if taup < h[i] for all values, iplane would still be -1 and we would be accessing an array with this as an index, which is undefined behavior.

In both cases, to avoid that, set that index to zero.

This was found using cppcheck tool.

[ymdatta]

I want someone familiar with the implementation details to review this and see if this is alright. Most of the variables changed here are internal variables and are specific to the implementation. Thanks!

[metzm]

The original code has been written in Fortran which uses 1 based indices. During translation to C, not everything was translated properly, particularly the adjusted of 1 based indices to zero-based indices. The correct fix would probably be to use k = i in https://github.com/OSGeo/grass/blob/main/imagery/i.atcorr/computations.cpp#L105 and kk = i in https://github.com/OSGeo/grass/blob/main/imagery/i.atcorr/computations.cpp#L112
Usage of kk then needs to be adjusted in https://github.com/OSGeo/grass/blob/main/imagery/i.atcorr/computations.cpp#L121 to for (i = kk; i < 83; i++) {

[ymdatta]

Made changes to reflect the suggestions.

[echoix]

I'll proxy the approval once it passes later tonight