[BUG] Numpy Unsatisfiable Dependencies using Python3.11+

**Describe the bug**
I am unable to deploy an unmodified agent using the starter toolkit on any python version higher than 3.10. When uv tries to build the dependencies it complains that there are no usable wheels that satisfy numpy==2.4.1. Switching to python3.10 solved the problem.

**To Reproduce**
Steps to reproduce the behavior:
1. Follow instructions on https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/agentcore-get-started-toolkit.html using python3.11 / 3.12 / 3.13
2. I used langgraph and all other recommended settings
3. Run `agentcore deploy`
4. See error

**Expected behavior**
No error and a successful build and deployment.

**Error Output**
```
⠹ Launching Bedrock AgentCore...Installing dependencies with uv for aarch64-manylinux_2_28...
❌ Failed to install dependencies with uv: Using CPython 3.11.13 interpreter at: /Users/joshuas/Projects/aws-bedrock-testing/.venv/bin/python3
× No solution found when resolving dependencies:
╰─▶ Because numpy==2.4.1 has no usable wheels and you require numpy==2.4.1, we can conclude that your requirements are unsatisfiable.

hint: Wheels are required for `numpy` because building from source is disabled for all packages (i.e., with `--no-build`)
```

**Environment:**
- OS: macOS 15.7.3
- Python version: error occurs on 3.11.13, 3.12.11, 3.13.5
- Package version: bedrock-agentcore-starter-toolkit 0.2.6
- Installation method: pip

**Additional context**
n/a

Start by following the linked AgentCore starter toolkit instructions with Python 3.11, 3.12, or 3.13 and run `agentcore deploy` to reproduce the uv resolution failure for numpy==2.4.1. Review how the toolkit constructs deployment dependencies and verify the result by achieving a successful build and deployment on those Python versions.

Written by the indexing model from the issue text.