Add: L4+ recursive Worker composition via _l3_worker_loop

[ChaoWao]

Summary

• Worker(level=4) registers Worker(level=3) children via add_worker(). On first run(), parent forks a child process per L3 child; the child inits the inner Worker, then enters _l3_worker_loop which polls the mailbox for (cid, config, args_blob), looks up the orch fn in the COW-inherited Python registry, and calls inner_worker.run(orch_fn, args, cfg).
• DistWorker::run() callback mechanism for THREAD mode: set_run_callback stores a Python callable; the binding layer acquires the GIL and reconstructs TaskArgs from TaskArgsView.
• _init_distributed / _start_distributed generalize the former _init_level3 / _start_level3 for all levels >= 3. DistWorker(self.level) replaces hardcoded DistWorker(3).
• read_args_from_blob(blob_ptr): nanobind binding to reconstruct TaskArgs from a mailbox blob.
• 13 new tests covering lifecycle, validation, single/multiple dispatches, L4 with own subs, multiple runs, and L3-owns-its-own-Orchestrator.
• Docs updated: task-flow.md §9 (L4→L3→L2 walkthrough), distributed_level_runtime.md §4, worker-manager.md §5.1 (nested fork ordering), roadmap.md (PR-F landed, PR-G includes ChipCallConfig → CallConfig).

Testing

• pip install . builds cleanly
• All 24 existing tests pass (no regressions)
• 13 new tests pass: test_l4_recursive.py
• grep -rn "DistWorker(3)" python/simpler/ returns 0
• Linux CI

[gemini-code-assist]

Code Review

This pull request implements L4+ recursive composition for the distributed runtime, allowing Worker instances to manage lower-level Worker children. Key changes include the introduction of a mailbox-polling loop for process-based recursion, support for thread-based recursion via GIL-acquired callbacks, and generalized initialization logic for all levels above L3. Feedback identifies a critical memory layout mismatch in mailbox offsets for ChipCallConfig and a performance issue where the worker polling loop busy-waits, consuming excessive CPU. Additionally, while the infrastructure for THREAD-mode recursion has been added, it is not yet fully integrated into the Worker class initialization logic.