Fixes an issue where indices aren't able to be created for tables in
schemas. At present, `vist_create_index` emits SQL like

```sql
CREATE INDEX index_name (col) ON schema_name.table_name
```
which results in an error when creating the table:
```
Index index_name cannot index a table/index schema_name.table_name
which is in a different named schema.
```
because the name needs to be prefixed by schema.

`vist_drop_index` does emit the correct SQL with the schema prefix.
```sql
DROP INDEX schema_name.index_name
```

This is due to an odd issue where the base class's
`visit_create_index` method has in its signature
`include_schema=False`:

https://github.com/sqlalchemy/sqlalchemy/blob/299284cec65076fd4c76bf1efaae60b60f4d4f7b/lib/sqlalchemy/sql/compiler.py#L6828C23-L6828C43

but `visit_drop_index` hard-codes the value to `True`:

https://github.com/sqlalchemy/sqlalchemy/blob/299284cec65076fd4c76bf1efaae60b60f4d4f7b/lib/sqlalchemy/sql/compiler.py#L6870

The dialects handle this by hard-coding `include_schema=True` in
`visit_create_index`, e.g.
sqlite:

https://github.com/sqlalchemy/sqlalchemy/blob/299284cec65076fd4c76bf1efaae60b60f4d4f7b/lib/sqlalchemy/dialects/sqlite/base.py#L1740

The difference in defaults is odd in the base class, but it seems like
include_schema=True is the appropriate setting for Spanner.