Meshroom nodes: replace documentation attribute with class docstrings and improve all node docs

[Copilot]

Recent Meshroom versions read node documentation from the class docstring (__doc__) rather than a documentation member variable. This PR migrates all nodes and takes the opportunity to substantially improve coverage and quality.

Description

Two changes are bundled:

1. Convert documentation attribute → class docstring across all 106 node files in meshroom/aliceVision/. Empty/whitespace-only documentation is dropped entirely.
2. Improve every node's docstring so each provides a meaningful multi-sentence description of purpose, behaviour, and typical usage context.

Before / After example (ConvertMesh.py):

# Before
class ConvertMesh(desc.AVCommandLineNode):
    commandLine = "aliceVision_convertMesh {allParams}"
    category = "Utils"
    documentation = """Convert a mesh to another mesh format."""

# After
class ConvertMesh(desc.AVCommandLineNode):
    """
Convert a 3D mesh from one file format to another.

Supported formats include OBJ, FBX, glTF, GLB, STL and PLY. The node can optionally flip
face normals (which may be required depending on the winding-order convention used by the
target application) and copy associated texture image files to the output folder.
"""
    commandLine = "aliceVision_convertMesh {allParams}"
    category = "Utils"

Features list

• Remove documentation class attribute from all 106 node files; replace with PEP-257 class docstring placed as the first statement in the class body.
• Add docstrings to 7 nodes that previously had none: ExportColoredPointCloud, ExportMatches, ImageMasking, LightingEstimation, SfMDistances, SfmBootstrapping, SfmExpanding.
• Expand 53 nodes whose docstrings were a single short line into multi-line descriptions covering purpose, algorithm, inputs/outputs context, and usage guidance.
• Fix prose issues in existing docs: typo form → from (SfMTransfer), of from → of (ConvertSfMFormat), de-indent internal-note style text (SfMPoseFlattening, SfMRigApplying, SfmLandmarksRemapping, MaskEroding).

Implementation remarks

The attribute-to-docstring migration was performed with a small Python script that:

• Detected documentation = """...""" / '''...''' patterns inside class bodies.
• Removed those lines (plus any trailing blank line).
• Inserted the content as a class docstring immediately after the class definition line.
• Skipped empty/whitespace-only values rather than emitting an empty docstring.

All 106 files pass python -m py_compile with zero errors.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[servantftransperfect]

Incorrect comment

[servantftransperfect]

Incorrect documentation (No morpho)

[servantftransperfect]

"suitable for panorama stitching" is theorically correct but misleading

[servantftransperfect]

It does not requires a panoramaRefining

[servantftransperfect]

No, PanoramaEstimation is not related to this node

[fabiencastan]

Suggested change

	This node scans the provided image files and directories, reads available metadata
	(e.g., EXIF focal length, sensor dimensions, serial number), and produces a minimal
	SfMData file listing all discovered views with their associated intrinsic parameters.
	The resulting SfMData is suitable as an input for the CameraInit node or other nodes
	that accept an SfMData file without requiring reconstructed poses.
	This node scans the specified files and directories to identify image files by their extension,
	then generates a minimal SfMData file listing. The output contains no metadata, intrinsics,
	or poses — making it ideal for pipelines such as the ImageProcessing node, where
	no knowledge of camera intrinsics or extrinsics is required.```

[fabiencastan]

Suggested change

	Given multiple directories of binary masks (where corresponding masks share the same filename),
	this node can combine them using logical operations (AND, OR, XOR) and apply morphological
	operations such as dilation and erosion. This is useful for refining or combining masks
	produced by different segmentation methods before using them in depth map estimation
	or texturing.
	Given multiple directories of binary masks (where corresponding masks share the same filename),
	this node can combine them using logical operations (AND, OR, XOR). This is useful for combining
	masks produced by different segmentations.