Add example of writing data back to Redshift using SQL API

This updates the README to include an example of saving data back to Redshift while using the SQL language API.

Fixes #81.

Author: Josh Rosen <joshrosen@databricks.com>

Closes #110 from JoshRosen/document-saving-data-from-sql.

README.md

@@ -115,14 +115,34 @@ df.write \
 
 #### SQL
 
+Reading data using SQL:
+
+```sql
+CREATE TABLE my_table
+USING com.databricks.spark.redshift
+OPTIONS (
+  dbtable 'my_table',
+  tempdir 's3n://path/for/temp/data',
+  url 'jdbc:redshift://redshifthost:5439/database?user=username&password=pass'
+);
+```
+
+Writing data using SQL:
+
 ```sql
+-- Create a new table, throwing an error if a table with the same name already exists:
 CREATE TABLE my_table
 USING com.databricks.spark.redshift
-OPTIONS (dbtable 'my_table',
-         tempdir 's3n://my_bucket/tmp',
-         url 'jdbc:redshift://host:port/db?user=username&password=pass');
+OPTIONS (
+  dbtable 'my_table',
+  tempdir 's3n://path/for/temp/data'
+  url 'jdbc:redshift://redshifthost:5439/database?user=username&password=pass'
+)
+AS SELECT * FROM tabletosave;
 ```
 
+Note that the SQL API only supports the creation of new tables and not overwriting or appending; this corresponds to the default save mode of the other language APIs.
+
 ### Hadoop InputFormat
 
 The library contains a Hadoop input format for Redshift tables unloaded with the ESCAPE option,
@@ -353,4 +373,4 @@ directly:
 
 Version 0.4+ adds the DataSource API and JDBC, which is an entirely new API, so although this won't break
 code using the InputFormat directly, you may wish to make use of the new functionality to avoid performing
-<tt>UNLOAD</tt> queries manually.
+<tt>UNLOAD</tt> queries manually.