[Optimization] Improve border/interpolation wrapper implementations

[zacharyvincze]

Description

This PR offers improvements to the interpolation, border paths as well as adding a dedicated WarpAffine kernel over using the WarpPerspective operator.

Optimizations to these paths improve geometric transformations which make use of border modes and interpolation (Rotate, Resize, WarpAffine, WarpPerspective, CopyMakeBorder).

Technical Details

• Uses device intrinsics where possible in border/interpolation wrapper implementations.
• Coordinate calculations use 32-bit integer math over 64-bit integer math when possible, but will fallback to 64-bit math when not possible.
• Interpolation implementations such as Bilinear and Bicubic now determine whether their neighborhood falls within image bounds. If they do, then they ignore the border calculations and sample pixels directly.
• WarpAffine now has a dedicated kernel instead of calling WarpPerspective. This ensures that affine warp doesn't use the bottom row of the 3x3 transformation matrix that a perspective transform uses, which was redundant anyway.

[Copilot]

Pull request overview

This PR refactors the affine warp path to use a dedicated WarpAffine kernel (instead of delegating to WarpPerspective) and optimizes border/interpolation wrappers with new shared sampling helpers to reduce overhead on common in-bounds sampling cases.

Changes:

• Added dedicated GPU/CPU WarpAffine kernels and updated WarpAffine::operator() to dispatch directly by dtype/channels/border/interp.
• Optimized InterpolationWrapper (bilinear/cubic) with in-bounds fast paths to bypass border handling when neighborhoods are fully in-range.
• Refactored border coordinate math and interpolation index conversion into a new sampling_helpers.hpp helper header.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
src/op_warp_affine.cpp	Replaces WarpPerspective reuse with dedicated WarpAffine dispatch and kernel launches.
include/op_warp_affine.hpp	Removes WarpPerspective member dependency; keeps WarpAffine API unchanged.
include/kernels/host/warp_affine_host.hpp	Adds host-side affine warp kernel implementation.
include/kernels/device/warp_affine_device.hpp	Adds device-side affine warp kernel implementation.
include/core/wrappers/interpolation_wrapper.hpp	Adds in-bounds fast paths and refactors nearest/bilinear/cubic sampling math.
include/core/wrappers/border_wrapper.hpp	Refactors border coordinate mapping with new helper functions and adds at_inbounds().
include/core/detail/sampling_helpers.hpp	New shared host/device helpers for modulo/clamp/abs and interpolation index conversion.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 59.47137% with 92 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
include/core/wrappers/interpolation_wrapper.hpp	34.43%	39 Missing and 1 partial ⚠️
src/op_warp_affine.cpp	75.82%	6 Missing and 16 partials ⚠️
include/core/detail/sampling_helpers.hpp	58.82%	14 Missing ⚠️
include/core/wrappers/border_wrapper.hpp	56.67%	11 Missing and 2 partials ⚠️
include/kernels/host/warp_affine_host.hpp	72.73%	0 Missing and 3 partials ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop     #147      +/-   ##
===========================================
- Coverage    74.40%   73.92%   -0.48%     
===========================================
  Files           79       81       +2     
  Lines         3355     3520     +165     
  Branches       738      764      +26     
===========================================
+ Hits          2496     2602     +106     
- Misses         378      416      +38     
- Partials       481      502      +21     

Files with missing lines	Coverage Δ
include/kernels/host/warp_affine_host.hpp	72.73% <72.73%> (ø)
include/core/wrappers/border_wrapper.hpp	71.43% <56.67%> (+6.61%)	⬆️
include/core/detail/sampling_helpers.hpp	58.82% <58.82%> (ø)
src/op_warp_affine.cpp	77.08% <75.82%> (-22.92%)	⬇️
include/core/wrappers/interpolation_wrapper.hpp	41.86% <34.43%> (-1.54%)	⬇️

... and 1 file with indirect coverage changes

🚀 New features to boost your workflow:

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