Support CREATE/DROP/ALTER VIEW (and associated DataFrame functions)

[Igosuki]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
Currently we can create external and in memory listing tables, I want to go further and allow creating views from SQL queries.
Registering a dataframe as a table requires the following :

    let table = ctx.clone().sql(&sql_query).await?;
    let df_impl = Arc::new(DataFrameImpl::new(ctx.state.clone(), &table.to_logical_plan()));
    ctx.register_table(table_name.as_str(), df_impl.clone())?;

Describe the solution you'd like
Since DataframeImpl is already a TableType::View, registering/droping/altering a view could be done from SQL, or directly with ctx.create_view(df: Arc<dyn DataFrame>) or df.register_as_view(name: impl Into<TableReference>)
and

ctx.sql("CREATE VIEW records AS select * from other_records where ...")

Describe alternatives you've considered
Leaving things as is

Additional context
Make it easier to use arrow-datafusion without knowing its internals.

[alamb]

🤔 since SchemaProvider already allows registering a new table provider, we might be able to implement this without any new traits

https://github.com/apache/arrow-datafusion/blob/6ec18bb4a53f684efd8d97443c55035eb37bda14/datafusion/src/catalog/schema.rs#L40-L51

[matthewmturner]

Im very interested in this as well, in particular just getting started with enabling CREATE VIEW.

@Igosuki did you have plans to work on this?

[Igosuki]

I would love to but I am quite busy with other things these days (recently got knee surgery). So feel free to take ownership !

[matthewmturner]

Sorry to hear - wishing you a speedy recovery!

I hope to look into it soon. Will keep you posted.

[alamb]

Create VIEW would be cool indeed -- I suggest adding it to the Catalog/Schema provider traits (as the view definitions logically belong there I think)

[matthewmturner]

@alamb noted - i will look into that.

[matthewmturner]

I have started work on CREATE VIEW