[ARCH] Replace heaviest traits in ClientHandler with injected collaborator objects

[jeromesinolanka]

Priority: 🟡 Medium

Description

ClientHandler currently composes 10 traits (see also issue #94 — MED-6). While that issue covers the immediate static analysis and maintainability concern, the deeper architectural goal is to move from implicit trait coupling to explicit dependency injection.

The three heaviest traits are the best candidates for extraction into standalone collaborator classes:

Trait	Lines	Why extract first
PerformsHttpRequests	1096	Already identified as a God Trait in CRIT-2 (#80)
ManagesPromises	~394	Self-contained async logic, minimal state dependencies
ManagesRetries	~247	Clean input/output contract, easy to mock in tests
ManagesConnectionPool	~unknown	Infrastructure concern, should be separately testable

With injected collaborators, ClientHandler becomes:

public function __construct(
    private RequestExecutorInterface $executor,
    private RetryManagerInterface $retryManager,
    private PromiseManagerInterface $promiseManager,
    private ConnectionPoolInterface $pool,
    // ...
)

This makes dependencies explicit, enables mocking individual concerns in tests, and eliminates the implicit shared-state problem.

Recommended Approach

This is a significant refactor best done in phases:

1. Phase 1: Extract ManagesPromises and ManagesRetries (fewer dependencies on handler state)
2. Phase 2: Extract ManagesConnectionPool
3. Phase 3: Extract PerformsHttpRequests (coordinate with CRIT-2 resolution)

Preserve the existing public API throughout — the extraction should be transparent to callers.

Related

Issue #94 — MED-6: God Class via 10 trait compositions
Issue #80 — CRIT-2: PerformsHttpRequests God Trait

[github-actions]

This issue has been automatically marked as stale due to inactivity.