AC: filters are applied on the original trace and not on a copy.

[ThomasLecocq]

This bug happens only when doing autocorrelation with more than one filter, and with filter bounds that are not "from the largest to the smallest" band.

So, for example:

1. Raw trace gets loaded
2. Filter 1 (0.1 - 1.0 Hz) is applied and AC is done
3. Filter 2 (1.0 - 2.0 Hz) is applied but on the previously already filtered trace = not much info left above 1.0 Hz
4. Filter 3 (0.1 - 2.0 Hz) is applied, again on the f2(f1(raw)), so only energy around 1.0Hz is remaining.

This Issue is just to explain the case (in the figures below, filter 1 label should be "0.1 - 1.0 Hz"):

With this (1ad6c9c) correction: