Channel ordering in NP2.4 recordings needs re-arranging

[oliche]

In the original order, shanks and rows are mixed. We were used to have non-contiguous shanks, but this is the first time we see this on the rows.

Example after sorting by shank, col, row:

The spike sorter has to operate on ordered channels, and the order used needs to be reproducible for all subsequent applications.

[oliche]

The implementation could solve this in the spikeglx.Reader object from ibl-neuropixel.
We would create a new header field ordered that orders by shank, row, col (default NP1.0 order), and apply by default.

[grg2rsr]

started to look at this: PR #48 in ibl-neuropixel