Fix incorrect logits returned in API

[mihow]

This PR fixes incorrect logits being returned in the API by restructuring how classifier results are handled and returned. The changes address issue #91 by ensuring logits are properly separated from scores in classification responses.

• Returns the actual logits from the model output rather than a copy of the softmax scores
• Introduces a new ClassifierResult dataclass to standardize classifier output format
• Add test coverage for expected logits and scores
• Fixes the test setup for all previous API tests that were passing because no images were in the tests requests.
• Fixes the test failures introduced in Remove algorithm and category map details in every result response #93 (but were passing because of the discovered test setup issue)

[sentry]

🔍 Existing Issues For Review

Your pull request is modifying functions with the following pre-existing issues:

📄 File: trapdata/api/models/classification.py

Function	Unhandled Issue
save_results	ValidationError: 2 validation errors for ClassificationResponse /pi... Event Count: 2

📄 File: trapdata/ml/models/base.py (Click to Expand)

Function	Unhandled Issue
run	ValidationError: 2 validation errors for ClassificationResponse /pi... Event Count: 2
run	NameError: name 'ceil' is not defined /process Event Count: 1

---

_{Did you find this useful? React with a 👍 or 👎}

[Copilot]

Pull Request Overview

This PR fixes incorrect logits being returned in the API by restructuring how classifier results are handled and returned. The changes address issue #91 by ensuring logits are properly separated from scores in classification responses.

• Introduces a new ClassifierResult dataclass to standardize classifier output format
• Updates classification processing to correctly handle logits vs scores
• Adds comprehensive test coverage for logits functionality

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
trapdata/ml/models/base.py	Adds ClassifierResult dataclass to standardize classifier output
trapdata/api/tests/test_api.py	Updates tests and adds new test for logits verification
trapdata/api/models/classification.py	Refactors post-processing to use ClassifierResult and fix logits handling

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[mihow]

Part of RolnickLab/antenna#782