python/sedonadb: Make explain plans easier to programatically inspect

In #733 and a few other places (I think Parquet pruning) we use the explain output to assert that something is or isn't happening in the internals of the engine. We currently don't have a great way to inspect these.

We should probably have `.explain()` return an object like:

```python
class ExplainPlan:

def __init__(self, df, options):
pass

def __repr__(self):
# Current output or close
pass
```

...to which we could add methods to more robustly inspect the structure/metrics.

The issue names no files or tests. Start by locating the Python `.explain()` implementation and tracing its current output; done should include an inspectable explain-plan object, methods for structure or metrics, and a representation that preserves the current output or stays close to it.

Written by the indexing model from the issue text.