Improvements for events, buffer is filling during backup compression

[thunderstorm]

I really appreciate that db compression is decoupled in the backup script, but it seems that events aren't necessarily executed and the buffer is filling - on my nano SD that takes about ~13min.
Would it be feasible to allow those updates during the compression / after the db is copied ? I'd assume db-locks aren't required anymore. Or is the event handling single-threaded or compression that expensive that other work should be delayed ?

[admin]

Events are serialized to avoid doing too much in parallel and causing out-of-memory issues by that.