Refactor label filtering logic

[blkt]

As per this comment, we currently allow filtering by label in two places (i.e. two endpoints), but the code to handle labels is duplicated.

In particular, the first implementation (here) handles input processing in the database layer, while the newer implementation has parameter splitting in the handler and inclusion/exclusion (i.e. label vs !label) in the service.

Refactor the older implementation into the second one, possibly reusing what is to be merged in the PR linked.