Update LlamaLanguageModel to replace implementation-specific properties with custom generation options

[mattt]

Follow up to #50

[Copilot]

Pull request overview

This PR refactors LlamaLanguageModel to move implementation-specific properties (contextSize, batchSize, threads, seed, temperature, topK, topP, repeatPenalty, repeatLastN) from model-level initialization parameters to per-request custom generation options. This follows up on #50 to improve the API design by allowing these parameters to vary per request rather than being fixed at model initialization.

Key changes:

• Properties previously set during initialization now default to standard values and can be overridden per request via GenerationOptions custom options
• Original initializer parameters are deprecated but remain functional (ignoring passed values)
• New internal ResolvedGenerationOptions struct consolidates option resolution logic
• Refactored sampling logic into a dedicated applySampling method

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
Sources/AnyLanguageModel/Models/LlamaLanguageModel.swift	Adds custom option properties to CustomGenerationOptions, deprecates model-level properties, introduces ResolvedGenerationOptions for option resolution, refactors initializers and generation methods to use resolved options, extracts sampling logic into applySampling method
Tests/AnyLanguageModelTests/LlamaLanguageModelTests.swift	Updates tests to reflect new initialization API, adds tests for custom options round-trip and deprecated initializer behavior, expands test for custom generation options with new parameters

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.