feat: add client.exists() or client.has_model() method to check if a model is available locally

Feature Request

Currently there is no clean way to check whether a model is already pulled locally without either:

1. Calling ollama.show(model) and catching the exception
2. Calling ollama.list() and iterating through all models

Proposed API

import ollama

if not ollama.exists("llama3.1:8b"):
    ollama.pull("llama3.1:8b")

Or on the client:

client = ollama.Client()
if not client.exists("llama3.1:8b"):
    client.pull("llama3.1:8b")

Use Case

When building applications that need to ensure a model is available before running inference (e.g., CI pipelines, first-run setup scripts, or tools like factory floor AI systems that need to work offline), a simple boolean check is much cleaner than exception-based flow control.

Current Workaround

try:
    ollama.show("llama3.1:8b")
except ollama.ResponseError:
    ollama.pull("llama3.1:8b")

This works but is not discoverable and uses exceptions for control flow, which is an anti-pattern in Python.

Start by tracing the existing show(), list(), pull(), and Client entry points to understand how local model availability is currently determined. Define the chosen module-level and client API consistently, then add coverage showing that an available model and a missing model produce the intended boolean result.