perf(auth): ⚡️ parallelize oauth credential initialization and deduplication

feat(setup): ensure proxy key validation runs after credential creation

feat(core): enable configuration of maximum concurrent requests per key

refactor(client): 🔨 make credentials optional and centralize validation

chore(config): ⚙️ enable conditional instantiation of API call persistence

refactor(core): review fixes + improve error handling and execution logic

fix(proxy_app): 🐛 add type key to aggregated tool calls initialization

feat(provider): add detailed transaction logging for gemini cli provider

fix(core): stabilize streaming for custom endpoints and eliminate initialization conflicts

feat(gemini-cli): implement tool use (function calling) and reasoning support

feat(core): introduce model whitelisting to override blacklists

refactor(auth): implement two-pass OAuth credential deduplication

refactor(auth): refine duplicate OAuth credential checking logic

feat(auth): introduce dedicated --add-credential CLI utility

feat(auth): implement automated credential discovery and duplication check

refactor(auth): enhance OAuth credential discovery and client logging

fix(auth): resolve concurrent interactive prompts and token renewal issues

feat(auth): implement robust OAuth credential initialization and management

feat(provider): introduce OAuth credential management and custom provider handling

feat(proxy): ✨ implement detailed per-request transaction logging

feat(api): ✨ adopt OpenAI schema and blacklist patterns for model discovery

fix(proxy): handle None values in streaming tool-call aggregation

feat(api): ✨ add CORS middleware to allow cross-origin requests

docs: restructure project licensing documentation

feat(client): enhance streaming error handling and log clarity

Fix(logging): enhance logging configuration and control

perf(litellm): improve litellm operational efficiency

feat(logging): centralize and enhance application logging

feat(rotator): allow explicit logging configuration for client

feat: Improve streaming error handling and refactor client completions

fix: Await response generator in chat completions for proper streaming

Refactor: Streaming response handling and key management.

feat: Improve streaming robustness and key management

feat: Abort API retries on client disconnection

fix(client): centralize model parameter conversion

feat(embeddings): Make server-side batching optional

feat(embeddings): Implement request batching for performance

feat(embeddings): Allow string input and update Pydantic calls

feat(proxy): Add granular litellm exception handling

feat: Add support for embeddings endpoint

refactor: Reconstruct complete response from stream for logging

feat: Add CLI args, lazy loading, and enhance Gemini provider

feat: Add release drafter configuration and refactor build process to use Python script

feat: Add build workflow, proxy application, and update documentation for executable usage

feat: Enhance request logging to ensure proper UTF-8 encoding in log files

feat: Implement request logging for API responses and enhance streaming response handling

feat: update proxy to use the new async context management

feat: Exclude PROXY_API_KEY from provider API key loading and enhance model retrieval options

feat(multi-provider): Implement dynamic API key loading and new endpoints

feat: Refactor project into reusable API key rotation library and proxy