Gmx_covar ouput to NxN matrix that can be used in python

GROMACS version: 2020.2
GROMACS modification: No

I am struggling with the output of gmx_covar. The -xpma option provides me a NxN covariance matrix as .xpm file. When I open this file (with gimp for instance), the diagonal does not equal 1 for all residues except the first and last. Is there any explanation for this observation?

Furthermore, I intend to use the covariance matrix to build a correlation matrix using python. Is there a convenient way to convert this photo-editor .xpm file to data file containing the the numbers for the “spots” rather than color-codes? gmx_xpm2ps does not yield anything useful either.
I assume there can be done some conversion of the covar.dat output-file, which does contain the matrix in 3NxN format. I don’t understand what the x, y and z columns in this file represent, could someone please explain? I am aware of this tool: Generalized Correlations | Max Planck Institute for Biophysical Chemistry, but unfortunately it has never been updated as far as I am aware.