API Conv prim

[MuawizChaudhary]

This PR adds the convolution primitive, which generalizes the scattering convolution operation. This PR will eventually resolve issue #305. Perhaps we can resolve issue #445 ?

[eickenberg]

Since this is rebased on another PR, it is a bit hard to review like this. Let's merge the rfft PR first, then rebase this on top. Then things will become clearer

[janden]

Same thing here. The history includes too many unrelated commits. Either rebase or I will review when the relevant PRs are merged.

[codecov-commenter]

Codecov Report

Merging #618 into dev will decrease coverage by 9.51%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##              dev     #618      +/-   ##
==========================================
- Coverage   88.38%   78.87%   -9.52%     
==========================================
  Files          64       65       +1     
  Lines        2325     2329       +4     
==========================================
- Hits         2055     1837     -218     
- Misses        270      492     +222     

Flag	Coverage Δ
#jenkins_main	?
#travis	78.87% <100.00%> (+0.12%)	:arrow_up:

Impacted Files	Coverage Δ
kymatio/scattering1d/frontend/numpy_frontend.py	66.66% <ø> (ø)
...matio/scattering1d/frontend/tensorflow_frontend.py	67.39% <ø> (ø)
kymatio/scattering1d/frontend/torch_frontend.py	92.85% <ø> (ø)
kymatio/scattering1d/convolution.py	100.00% <100.00%> (ø)
kymatio/scattering1d/core/scattering1d.py	95.23% <100.00%> (-0.60%)	:arrow_down:
kymatio/scattering1d/frontend/base_frontend.py	100.00% <100.00%> (ø)
kymatio/backend/torch_skcuda_backend.py	0.00% <0.00%> (-100.00%)	:arrow_down:
...matio/scattering1d/backend/torch_skcuda_backend.py	0.00% <0.00%> (-100.00%)	:arrow_down:
...matio/scattering2d/backend/torch_skcuda_backend.py	0.00% <0.00%> (-100.00%)	:arrow_down:
...matio/scattering3d/backend/torch_skcuda_backend.py	0.00% <0.00%> (-85.19%)	:arrow_down:
... and 5 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 4f65e51...9f72396. Read the comment docs.

[MuawizChaudhary]

What will tests for the convolution primitive look like?