Implement `ShadeClient` class for per-instance configuration

[codebestia]

Description:

While global config is convenient, multi-tenant applications (e.g. a SaaS acting on behalf of multiple merchants) need isolated client instances that carry their own credentials. ShadeClient should accept the same parameters as the global config and bind them to a single instance, which is then passed to resource methods.

Proposed Steps:

• Create src/shade/client.py with a ShadeClient dataclass/class.
• Constructor parameters: api_key, environment, api_base, timeout, max_retries.
• Each resource class should accept an optional client= kwarg and fall back to global config if omitted.
• Provide a ShadeClient.from_env() factory that reads SHADE_API_KEY and SHADE_ENVIRONMENT from environment variables.

Acceptance Criteria:

• client = ShadeClient(api_key="sk_test_xxx") creates an isolated client.
• Resource calls using this client use its credentials, not global config.
• Two ShadeClient instances with different keys can coexist without interfering.
• ShadeClient.from_env() reads SHADE_API_KEY from the environment correctly.
• Missing api_key on the instance when global is also None raises AuthenticationError.

[Whiznificent]

gm maintainer i will like to work on this issue

[Whiznificent]

gm maintainer i can fix this issue and have the pr submitted to you by tomorrow asap

[stiven-skyward]

Hi! As a Cloud Engineer and architect who frequently works with Python, I have extensive experience building scalable client SDKs. I understand the importance of isolating state, and I can implement the ShadeClient class to ensure clean, per-instance configuration that avoids global state leaks. I can submit a solid PR within 24 hours. Best, Stiven.