Logging time series metrics to an experiment run with step=0 goes the wrong way

#### Environment details

- OS type and version: Linux
- Python version: `all`
- pip version: `all`
- `google-cloud-aiplatform` version: `main`

#### Steps to reproduce

1. call `ExperimentRun.log_time_series_metrics()` with `step=0`.

The main issue here is the condition at [this line](https://github.com/googleapis/python-aiplatform/blob/main/google/cloud/aiplatform/metadata/experiment_run_resource.py#L955):

`if not step:`

Should be:

`if step is None:`

A caller that counts its steps starting at 0, would unnecessarily start a search for the latest time series step and that could trigger other issues, for example, hitting a limit on the size of the time series to read, which is not relevant here, but could be avoided if this condition was testing the right value.

#### Code example

```python
run = aiplatform.ExperimentRun(experiment=my_experiment, run_name=my_run_name)
run.log_time_series_metrics(my_metrics, step=0)
```

#### Stack trace

Start at google/cloud/aiplatform/metadata/experiment_run_resource.py around line 955 and trace ExperimentRun.log_time_series_metrics(). Verify that the provided step=0 follows the explicit-step path rather than the missing-step path; done when logging at step 0 no longer searches for the latest time-series step.

由索引模型根据 Issue 内容生成。