reassign single flags column

in context of harmonization and dataprocessing the flags get reshaped. To avoid direct overriding of flagger._flags there is currently this workaround in use:

flags = flagger.getFlags()
flags[field] = new_flags
flagger_out = flagger.initFlags(flags=new_flags)  

The problem with this attempt is, that, when reinitializing the whole flags frame at the end, all eventually present additional information - such as dmp flaggers cause/comment field gets lost for all variables.

When i get it right, currently its not possible to achieve reshaping of a flags column of a flagger, without accessing its ._flags attribute or overriding additional, flaggger specific information.

I could think of a column specific flags reshaping method of the flagger, or of giving .setFlags the extra possibillity of initializing a flags column that is not already present - so one could drop a column-to-be-reshaped with a simple drop method and than reinitialize it with setFlags and a new flags column.

But since you @palmb and @schaefed are a bit more familiar with in the flaggers dos-and-donts, i thought its best one of you suggests/discusses a solution.