REF: Remove padding and upadding operations from usfft

Purpose

Reduce the memory footprint of USFFT operations. Closes #66

Approach

Padding and unpadding operations in the usfft were using "wrap" mode that were associated with the scatter gather operations. These padding operations were equivalent to using modular arithmetic to wrap the indices of these operations. By using special indexing instead of padding and unpadding we remove additional memory usage that would grow with the desired accuracy of the transform.

Benchmark with the Laminography adjoint operator test: n=224, ntheta=224, tilt=np.pi/3 where "used memory" is the volume of all arrays allocated during the test and "aquired memory" is the maximum size of the memory pool at any time.

Before:	eps	used memory
1e-2	8.70GB	3.40GB
1e-4	8.78GB	4.95GB
1e-6	8.86GB	5.01GB

After:	eps	used memory
1e-2	7.24GB	2.01GB
1e-4	7.24GB	2.01GB
1e-6	7.24GB	2.01GB
1e-9	7.24GB	2.01GB

It seems that the change in memory footprint due to changing the accuracy of the transform is now negligible.

Pre-Merge Checklists

Submitter

[x] Write a helpfully descriptive pull request title.
[x] Organize changes into logically grouped commits with descriptive commit messages.
[x] Document all new functions.
[x] Write tests for new functions or explain why they are not needed.
[ ] Build the documentation successfully
[x] Use yapf to format python code.

Reviewer

[ ] Actually read all of the code.
[ ] Run the new code yourself.
[ ] Write a summary of the changes as you understand them.
[ ] Thank the submitter.

AdvancedPhotonSource / tike