Introduce the GCPDefinition

[mraspaud]

Add a new class called GCPDefinition (inherits from SwathDefinition) which provides access to the underlying gcps.

Use cases:

• faster loading of coarse lon/lats (for the boundary)

• if you want to write data with tiepoints

• interpolate a fraction of what is needed.

• Closes #xxxx

• Tests added

• Tests passed

• Passes git diff origin/main **/*py | flake8 --diff

• Fully documented

[mraspaud]

@BENR0 ^

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.29%. Comparing base (1268d90) to head (71a019c).
⚠️ Report is 774 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #476      +/-   ##
==========================================
+ Coverage   94.28%   94.29%   +0.01%     
==========================================
  Files          69       70       +1     
  Lines       12394    12417      +23     
==========================================
+ Hits        11686    11709      +23     
  Misses        708      708              

Flag	Coverage Δ
unittests	94.29% <100.00%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[coveralls]

Coverage increased (+0.01%) to 93.802% when pulling 71a019c on mraspaud:feature-tiepoint-def into 1268d90 on pytroll:main.

[djhoese]

I'm not sure I understand all of the use cases for GCPs, but I thought they were a sparse version representing something much larger. How does this work here with your definition? Put another way, I thought the user would give GCPs and like an overall shape of the area and that would be used to generate the full lons/lats on the fly. Where is get_coarse_bbox_lonlats going to be used? It isn't used currently, right? How useful is it when we have access to the full lons/lats already? How will the user control whether "coarse" bounding boxes are good enough for their use case?

[mraspaud]

I'm not sure I understand all of the use cases for GCPs, but I thought they were a sparse version representing something much larger. How does this work here with your definition? Put another way, I thought the user would give GCPs and like an overall shape of the area and that would be used to generate the full lons/lats on the fly. Where is get_coarse_bbox_lonlats going to be used? It isn't used currently, right? How useful is it when we have access to the full lons/lats already? How will the user control whether "coarse" bounding boxes are good enough for their use case?

Yes, very good questions indeed.

1. This is just a proof of concept when testing SwathDefinition html representations (for speed) so it's not really used anywhere yet.
2. Indeed, we could make the GCPDefinition have interpolation functions. However a problem is how would it know what interpolation scheme to use? One solution would be a callback to the interpolation function for example.
3. The coarse bbox is basically just a quick test for now. I have no idea how we could quantify the quality of that, or fitness to give use case...

[BENR0]

I think another use case that was mentioned by @mraspaud was to be able to write geotiffs with only tiepoints.