Incorrect encoding of numpy 1D vectors

this issue was reported by Gavin Lobo, the following commands failed to return the correct array length using bjdata v0.4.0

```python
import bjdata as bj
import numpy as np
bj.dumpb(np.array([1,2,3,4],dtype=np.int8))
    b'[$i#U\x01\x01\x02\x03\x04'
```

where the array length `#U\x01` was incorrect, should have been `#U\x04` as there are 4 elements.