Fixes for normalization issues

[AsmaTANABEN]

Adjust normalization of density compensation coefficients and adding Cufinufft Pipe function authored by Chaithya G.R.
Resolves #211

[paquiteau]

Ideally we have some tests as well ?

[chaithyagr]

I would strongly recommend adding the changes for cufinufft in #195 as it is tested there. Also, that will allow this PR to be merged smoothly without waiting for finufft/cufinufft releases.

[paquiteau]

Questions for you: can this normalization rule be used for the other density compensations methods (e.g. voronoi) ?

[chaithyagr]

Questions for you: can this normalization rule be used for the other density compensations methods (e.g. voronoi) ?

Answer for you: Yes :). It gives stable results.

[chaithyagr]

LGTM, assuming it matches same as tfkbnufft and tensorflow-nufft. I will rely on you for it @AsmaTANABEN , I dont think we need those tests here run regularly.

[chaithyagr]

I think we can merge when green and @paquiteau lets make a release?

[chaithyagr]

I think this can be merged right?

[paquiteau]

We could merge it yes. However I think this normalization should be refactored to be generalizable to the other density compensation methods (e.g. voronoi, cell-counting etc.). If this is blocking we can post-pone this refactoring to another PR.