Seattle rig consistently failing QC

[k1o0]

Sessions have QC outcomes such as this. The frame2ttl has been calibration and the issue lies with the Bonsai workflow. This needs to be investigated (separate issue) and the extraction should be modified to ignore consistent sync square changes before stim on (see #654).

There is also an issue with the Harp sound card (see #655). It seems particularly bad on this rig so we should investigate if the TTLs are indeed failing in the same manner, or whether it is a wiring issue. Checking Bpod and DAQ TTLs in the QC viewer will tell us more information.

[oliche]

The QC Viewer can't be run on NP2.4
This is due to the dynamic pipeline creation that attempts to recreate a pipeline after the splitting of the original probes in 4 shanks.

Here is a quick fix to address this at the QC stage, which is not desirable @mayofaulkner @k1o0
https://github.com/int-brain-lab/ibllib/compare/taskqc?expand=1

[oliche]

This issue underwent 2 phases of mitosis and is now split in 4 issues

1. Harp sound card fails to output TTLs #655 Harp
2. int-brain-lab The QC Viewer can't be run on NP2.4 ibllib#768 task qc with np2.4
3. ephysCW v8 extracts with the wrong sequence Ephys Choice world is extracting with the wrong pre-defined sequence on V8 #658
4. frame2TTL extraction and unexpected flicker before stimOn (probably due to ITI definition?) Estimate impact of reverting the stimOn extraction to first pulse after stimOnTrigger #654