New roboflow model block versions with confidence dropdown

[leeclemnet]

What does this PR do?

#2242 introduced support for "best", "default" and custom confidence threshold modes with "best" derived from model-eval and "default" using the inference_models defaults / env vars. "default" is the default and inference users generally must opt-in to the the other modes.

This PR creates new versions of the roboflow model workflows blocks to include some UI around confidence mode selection, and makes "best" the recommended and default option for new workflows.

Semantic Segmentation and Keypoint Detection model blocks are updated to match the UI but do not have the "Best" option because model-eval does not yet support those tasks.

Depends on:

• https://github.com/roboflow/roboflow/pull/11053
• Support optimal confidence from model-eval #2242

Type of Change

• New feature (non-breaking change that adds functionality)

Testing

• I have tested this change locally
• I have added/updated tests for this change

Test details:

• Unit tests for new block versions

• Bump block versions in integration tests

• Workflow test against local inference server.

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code where necessary, particularly in hard-to-understand areas
• My changes generate no new warnings or errors
• I have updated the documentation accordingly (if applicable)

Additional Context