refactor: reduce complexity of process_agent_memory in cache_memory/execute.rs

[github-actions]

What was complex

process_agent_memory in src/tools/cache_memory/execute.rs had a Clippy cognitive complexity score of 33/25. The main loop processed each file with six sequential guard checks (symlink-containment, path safety, extension filter, size budget, content scan, copy) each using continue to skip files — deeply nested, hard to follow, and difficult to test in isolation.

What changed

Extracted process_memory_file helper with a FileOutcome enum:

enum FileOutcome {
    Copied(u64),
    Skipped(String),
}

async fn process_memory_file(
    relative_path, source_file, canonical_base, dest_file, config, total_size
) -> Result<FileOutcome>

Each guard in the helper uses an early return (return Ok(FileOutcome::Skipped(reason))) instead of a continue in the outer loop. This eliminates all the nesting complexity at the call site.

The main loop in process_agent_memory collapses to a clean match:

match process_memory_file(...).await? {
    FileOutcome::Copied(size) => { total_size += size; copied_count += 1; }
    FileOutcome::Skipped(reason) => { skipped_count += 1; skipped_reasons.push(reason); }
}

Additionally removed the now-unnecessary if !skipped_reasons.is_empty() guard before logging (iterating an empty Vec is a no-op).

Before / After complexity

Metric	Before	After
Clippy cognitive complexity	33/25 ⚠️	below 25 ✅

Verification

• All 87+ tests pass unchanged
• cargo clippy --all-targets --all-features -W clippy::cognitive_complexity reports zero cognitive-complexity warnings
• No behaviour changes — identical validation logic, same skip/copy outcomes

Generated by Cyclomatic Complexity Reducer · ● 804K · ◷