Some CPF fields are not worth us including

Some of the CPF columns are not worth us having in the metadata. It would be much cleaner if we dropped a bunch of these.

Fore example - all the columns that are just filled with `None` can be found using:
```python
df = pd.read_parquet(f"{URL}/parquet/shots")
df.columns[df.isna().all()].tolist()
```
I suggest we drop these for a start.