Support configurable worktree paths per project

### Feature summary

Allow custom worktree location and naming per project

### What problem are you trying to solve?

Today, worktrees are created in a fixed structure like:

```text
repo_clone_path/
└── copilot-worktrees/
└── repo_clone_name/
└── random-worktree-name
```
For some projects, especially large C++ codebases, this can create very long filesystem paths. Those long paths can cause build or compilation issues or make the worktree harder to use in environments with strict path-length limits.

### Proposed solution

Allow users to customize the worktree location and naming structure on a per-project basis.

Possible improvements:

- Let users set a custom base worktree directory
- Let users customize the worktree folder structure
- Prefer shorter or more predictable generated worktree names
- Optionally support random-name generation patterns that keep paths compact

### Workflow impact

This would improve reliability for projects that are sensitive to path length, especially C++ and other build-heavy repositories. It would also make worktrees easier to manage for users who need a more compact or organization-specific directory layout.

### Installation context

Project-level worktree setup in the GitHub App

### Additional context

Current structure:
```text
repo_clone_path/
└── copilot-worktrees/
└── repo_clone_name/
└── random-worktree-name
```
A customizable structure would help avoid long path issues and make worktrees more accessible in constrained environments.

No se nombran archivos, pruebas ni puntos de entrada de implementación. Empieza por localizar la configuración del worktree a nivel de proyecto de la GitHub App y seguir cómo produce la estructura actual repo_clone_path/copilot-worktrees/repo_clone_name/random-worktree-name. Antes de la implementación, aclara qué opciones de personalización y reglas de nomenclatura son necesarias; el trabajo debería incluir una configuración por proyecto que produzca de forma fiable rutas de worktree más cortas y predecibles.

Escrito por el modelo de indexación a partir del texto del issue.