Ignore cached analysis results based on criteria

[michaelweiser]

We've long been toying with the idea of re-analysing samples when cached analysis results get too old. Up until now we've equated this with deleting cached results from the database based on their timestamp or other criteria, which is a valid approach. While actually more of a maintenance command to not let the database grow indefinitely, it indirectly causes re-analysis of known samples.

Another idea to make this more explicit is to ignore cached results based on configurable criteria similar/in extension to #153, maybe as part of an expression.

We'll still need the database maintenance functionality though, e.g. as part of peekaboo-util (#154).