feat(geomltoolkits): integration with geomltoolkits for modularization

[kshitijrajsharma]

What does this PR do ?

• PR integrates new post processing technique for the vectorization using https://github.com/kshitijrajsharma/geoml-toolkits
• Makes code modular with integration of geoml-toolkits
• Removes legacy modules

[Copilot]

Pull Request Overview

This PR integrates geomltoolkits for modularizing prediction-related functionality and removes legacy modules.

• Updates pyproject.toml to include geomltoolkits and switch to Poetry for dependency management.
• Removes deprecated modules (downloader.py, georeferencer.py, raster2polygon.py, vectorizer.py) and updates predictor/app.py to use geomltoolkits functions.
• Adapts API endpoints and tests to asynchronous execution and streamlines model prediction functions.

Reviewed Changes

Copilot reviewed 16 out of 19 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
pyproject.toml	Updates project metadata and dependencies including geomltoolkits.
predictor/utils.py	Refactors imports and adds georeference_prediction_tiles implementation.
API/main.py	Updates API configuration and middleware; adds asynchronous endpoints.
predictor/prediction.py	Refactors prediction functions to use new toolkit functions and adjusts tensor extraction.
predictor/app.py	Updates prediction workflow to use geomltoolkits for tile downloading and georeferencing.
Various test and GitHub Actions	Updates workflows and tests for Poetry and async execution; removes legacy build files.

Files not reviewed (3)

• API/requirements.txt: Language not supported
• Dockerfile: Language not supported
• yolo-requirements.txt: Language not supported

Comments suppressed due to low confidence (1)

predictor/prediction.py:193

• The variable 'image_batch' is undefined in the 'keras' branch; replace it with 'image_paths' to ensure the correct set of images is processed.

    predict_keras(model, image_batch, confidence)