We only mask 'bad' values, what about 'suspicious' ones?

During the discussion in #61 (closed) I realized, that we never talked about the handling of suspicious flags and AFAIK all our tests simply ignore them. If we keep the masking of flagged values (what I favor), we probably should come up with a way to have fine grained control over this behavior.

Maybe through a writable flagger property, indicating the last acceptable flag? So the masking could be rewritten as something like that:

mask = flagger.isFlagged(flag=flagger.VALID, comparator=">=")
data_in = data.copy()
data_in[mask] = np.nan

To be consistent we should then also use flagger.VALID (or however we want to call such a property) as a default in setFlags, getFlags and so on. Any opinions on that?