mlxlm: MLXLMTypeAdapter.format_output_type uses falsy check instead of identity-None check

[harsh4vardhan]

MLXLMTypeAdapter.format_output_type (line 97) uses if not output_type instead of if output_type is None:

# mlxlm.py:97
def format_output_type(self, output_type=None):
    if not output_type:   # wrong: discards any processor where bool(processor) is False
        return None
    return [output_type]

If a logits processor subclass defines __bool__ returning False, the processor is silently discarded and generation proceeds unconstrained — no error raised. Every other model file uses the correct pattern (transformers.py, vllm.py, llamacpp.py, etc.).

Minimum example:

class FalsyProcessor:
    def __bool__(self): return False

def format_output_type_buggy(output_type=None):
    if not output_type:       # exact mlxlm.py code
        return None
    return [output_type]

p = FalsyProcessor()
print(p is None)                        # False
print(format_output_type_buggy(p))     # None — processor silently dropped

Fix: replace if not output_type: with if output_type is None:, consistent with every other model backend.