Scipy has a peak-finding routine that looks interesting - find_peaks that was added in 1.1.0. It would be worth testing whether it is faster than EQcorrscan's current routine (or similarly fast - maintaining things is effort), and gets the same results. It also looks like it could be useful for spike detection,

Very small suggestion;

Change the flags in [noedits] to TRUE only after successful completion.

Context:
If the filtering section breaks due to an error (e.g. not enough diskspace), no file is created, but the [noedits] section has already been set to TRUE, so that restarting will given an error (params file says that there is a filtered file, but there is not).

If you choose a long file name, it doesn't keep the form centered: