Fix ZeRO stage to choose BF16 optimizer in test (deepspeedai#7803)

Use ZeRO stage 1 to use BF16 optimizer. 
(We should have switched to ZeRO1 in deepspeedai#7788, but I missed the change.
@sfc-gh-truwase)

- deepspeedai#7790 removed the fallback that allowed bf16 model + fp32 grad
accumulation without ZeRO, so that combo now raises NotImplementedError.
- deepspeedai#7788 changed test_bf16_optimizer_fragments to force BF16_Optimizer by
setting grad_accum_dtype=fp32, but it kept ZeRO stage 0, which is now
invalid after deepspeedai#7790.

Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com>

Author: tohtana

tests/unit/runtime/zero/test_zero_tensor_fragment.py

@@ -179,7 +179,7 @@ def test_bf16_optimizer_fragments(self, frozen_weights):
                 "grad_accum_dtype": "fp32"
             },
             "zero_optimization": {
-                "stage": 0,
+                "stage": 1,
             }
         }