fix: raise TimeoutError on ZMQ retry exhaustion instead of returning None

[GaneshPatil7517]

Summary

Fixes #393 — recv_json_with_retry() returned None after 5 failed attempts, which propagated
to user control loops causing TypeError crashes at runtime.

Changes

concore_base.py

• recv_json_with_retry() — now raises TimeoutError instead of returning None
• send_json_with_retry() — now raises TimeoutError instead of silently returning None
• read() — added explicit except TimeoutError handler that returns default_return_val
• write() — added explicit except TimeoutError handler that logs the error without crashing

tests/test_concore.py

• Added TestZMQRetryExhaustion class with 4 tests:
  • recv_json_with_retry raises TimeoutError after 5 failed attempts
  • send_json_with_retry raises TimeoutError after 5 failed attempts
  • read() returns default_return_val on ZMQ timeout (not None)
  • write() handles send timeout gracefully without crashing

tests/test_concoredocker.py

• Added TestZMQRetryExhaustion class with 2 tests covering the same timeout paths via concoredocker

How it works

Previously, exhausted retries returned None silently — no exception was raised, so the existing
except blocks in read()/write() never triggered. Now TimeoutError is raised, which is caught
by the new explicit handler (and would also be caught by the existing except Exception fallback).
This ensures read() always returns a usable value (default_return_val) and write() never
crashes the caller.

Test results

All 60 tests pass (test_concore.py + test_concoredocker.py + test_protocol_conformance.py).

Fixes #393