chore: Add models for output and improve typing

[marc-at-brightnight]

• Add pydantic models for outputs
• Improve typing
• Lock poetry file
• Fix imports

This PR focuses on improving typing safety in core parts of the repo. Specifically, the introduction of pydantic models adds a source of truth, and removes the need to .get everything.

The output is slightly changed. Previously, we had:

ANALYST SIGNALS:
+-------------------+----------+--------------------+
| Analyst           |  Signal  |         Confidence |
+===================+==========+====================+
| Fundamentals      | BULLISH  |              75.0% |
+-------------------+----------+--------------------+
| Technical Analyst | NEUTRAL  |                12% |
+-------------------+----------+--------------------+
| Valuation         | BEARISH  | 56.00000000000001% |
+-------------------+----------+--------------------+
| Sentiment         | BEARISH  |              85.0% |
+-------------------+----------+--------------------+
| Risk Management   |          |              None% |
+-------------------+----------+--------------------+

Notice the last row is always blank. Now:

ANALYST SIGNALS:
+-------------------+----------+--------------+
| Analyst           |  Signal  |   Confidence |
+===================+==========+==============+
| Fundamentals      | BULLISH  |        75.0% |
+-------------------+----------+--------------+
| Sentiment         | BEARISH  |        85.0% |
+-------------------+----------+--------------+
| Technical Analyst | NEUTRAL  |        12.0% |
+-------------------+----------+--------------+
| Valuation         | BEARISH  |        56.0% |
+-------------------+----------+--------------+

This also aligns better with the CLI user selection.