ttv return value type consistency

[dmdunla]

Currently, we have the following output types for ttv in the various data classes:

- ktensor.ttv -> Union[float, ktensor]
- tensor.ttv -> tensor
- sptensor.ttv -> Union[sptensor, ttb.tensor]
- ttensor.ttv -> Union[float, ttensor]

All should support float and I think they do. But this should be verified and the type hints should be updated to include float across all.