Introduce scan_size for overlapping scans

[mraspaud]

This PR adds support for modis and viirs swath resampling with gradient search

• Closes #xxxx
• Tests added
• Tests passed
• Passes git diff origin/master **/*py | flake8 --diff
• Fully documented

[coveralls]

Coverage decreased (-0.1%) to 91.135% when pulling c679eb4 on mraspaud:feature-npp-gradient into f710309 on pytroll:master.

[codecov]

Codecov Report

Merging #248 into master will decrease coverage by 0.11%.
The diff coverage is 21.42%.

@@            Coverage Diff             @@
##           master     #248      +/-   ##
==========================================
- Coverage   91.27%   91.15%   -0.12%     
==========================================
  Files          42       42              
  Lines        8243     8257      +14     
==========================================
+ Hits         7524     7527       +3     
- Misses        719      730      +11

Impacted Files	Coverage Δ
pyresample/gradient/__init__.py	86.17% <21.42%> (-8.32%)	⬇️

---

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 f710309...c679eb4. Read the comment docs.

[djhoese]

I believe the viirs and modis readers in Satpy have a "rows_per_scan" piece of metadata that I use for EWA resampling when available. I'm willing to switch it, but I think I'd argue that rows_per_scan is more explicit.

[djhoese]

@mraspaud Any updates or thoughts on this PR after all your work on the other PRs? As I mentioned above there is rows_per_scan in the metadata for Satpy's VIIRS and MODIS readers.