bug: orderbook refresh thread crashes silently when API fetch fails

[skyc1e]

The background orderbook refresh thread in OrderBookManager._thread_refresh_order_book crashes permanently if _run_get_orders() returns None (e.g. API timeout or 429).

Root cause

_run_get_orders() catches exceptions and returns None on failure (line 333). The state update correctly guards with if orders is not None (line 377). But the debug log immediately after does not:

# orderbook.py, line 388-392
self.logger.debug(
    f"Fetched the order book"
    f" (orders: {[order.id for order in orders]}, "  # TypeError if orders is None
    ...
)

This throws TypeError: 'NoneType' object is not iterable. The except block at line 393 only catches ValueError, so the TypeError propagates and kills the thread.

Impact

After a single transient API failure:

1. The daemon refresh thread dies silently (no log, no restart)
2. _state is never updated again
3. The bot continues operating with a stale orderbook indefinitely
4. Stale prices/positions lead to incorrect order placement

This is particularly easy to trigger with the rate limiting described in #78.

Reproduction

1. Start the market maker
2. Temporarily block or rate-limit the CLOB API (e.g. firewall rule, or wait for a 429)
3. _run_get_orders() returns None
4. Debug log crashes the thread
5. No further orderbook refreshes happen

Fix

Either guard the log:

if orders is not None:
    self.logger.debug(
        f"Fetched the order book"
        f" (orders: {[order.id for order in orders]}, "
        ...
    )

Or broaden the exception handler from except ValueError to except Exception so transient failures don't kill the thread.