Release 2024.10.01

Fixed issues:

• multiple object labels were occasionally classified in a single patch; this is no longer possible
• occasional module import errors that differed from development/test/production environments. Importantly, scripts now sit outside of model_server package.

Set up configuration to build model_server as a conda package; see README.md for near-term package installation instructions

Accessor and PatchStackAccesor data models: added convenience functions for data access and pipelining array computations

RoiSet data structure: added support for RoiSets to be used in object detection tasks

• created distinct generator class methods for RoiSet
• RoiSet.from_bounding_boxes can create RoiSet with just YXWH coordinates and no segmentation, e.g. as output from object detection models
• can also inspect and quantify overlap between and within RoiSet(s), at both bounding box and segmentation level
• can generate RoiSet from polygon segmentation e.g. in COCO segmentation models

Removed loosely structured workflows. Refactored existing base and ilastik "workflows" as pipeline modules, e.g. base.pipelines.segment, with the following blocks:

1. input parameters (Pydantic model)
2. output record (Pydantic model)
3. FastAPI mapping and docstring; also any special input validation that doesn't play well Pydantic, and special exception routing
4. pipeline logic function, which pushes results onto PipelineTrace

Moved sample ImageJ client to trec-adaptive-feedback project. This now pulls from tagged releases, not the master branch, of model_server. So no pre-release integration testing was performed with the TREC codebase.