Refactor ActionsSubtask for more precise initialization logic

[collindutter]

• I have read and agree to the contributing guidelines.

Describe your changes

Fixed

• Answer: being trimmed from LLM's final answer even when using native tool calling.

This PR changes the logic of ActionsSubtask to be more precise about how it parses the input Actions. Previously, it followed the decision tree of:

• if TextArtifact: Parse ReAct prompt generated from Prompt Driver with use_native_tools=False.
• else: Parse Artifact generated from Prompt Driver with use_native_tools=True.

The problem is that even Prompt Driver with use_native_tools=True eventually generates a TextArtifact. Now the decision tree looks like:

• if TextArtifact and the TextArtifact has been explicitly marked as a ReAct Prompt: Parse ReAct prompt generated from Prompt Driver with use_native_tools=False.
• else: Parse Artifact generated from Prompt Driver with use_native_tools=True.

The way ReAct Artifacts are "marked" is via a metadata field, is_react_prompt. Technically this change of behavior is a breaking change but I highly doubt anyone is passing inputs into ActionsSubtask directly.

Issue ticket number and link

NA

[codecov]

Codecov Report

Attention: Patch coverage is 95.83333% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
griptape/tasks/tool_task.py	50.00%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

[dylanholmes]

Nice work! -- Just one subjective comment