[get_df] Updating multi-statement logic

[john-bodley]

This PR fixes an issue we've been running into with multi-statements in Presto which is asynchronous. When using the cursor associated with a raw connection using multiple statements, i.e.,

cursor = engine.raw_connection().cursor()
cursor.execute(...)
cursor.execute(...)

for Presto it wouldn't poll to check that the first statement had finished before executing the second. It seems that some DB-API's have the nextset method (though this isn't universal) but reading through code snippets it seems the best way to ensure that multi-statements execute sequentially for all engine types is to simply call fetchall from the cursor which waits until the statement is done, i.e., here's the logic for Presto.

The other option is to call db_engine_spec.handle_cursor after every statement (which is what SQL Lab does) which polls when necessary, however this logic is very much configured for SQL Lab queries.

Note that when using the sqlalchemy.engine.base.Connection (without a cursor) the following would works. I believe the reason is because execute returns a result set which enforces syncronicity:

conn = engine.connect()
conn.execute(...)
conn.execute(...)

However this approach isn't viable as we need to use the cursor to execute queries via db_engine_spec.execute.

to: @graceguo-supercat @michellethomas @mistercrunch @timifasubaa

[codecov-io]

Codecov Report

Merging #5517 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #5517      +/-   ##
==========================================
+ Coverage   63.28%   63.28%   +<.01%     
==========================================
  Files         349      349              
  Lines       22121    22123       +2     
  Branches     2457     2457              
==========================================
+ Hits        13999    14001       +2     
  Misses       8108     8108              
  Partials       14       14

Impacted Files	Coverage Δ
superset/models/core.py	87% <100%> (+0.04%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3b6cafc...d3a1de6. Read the comment docs.

[michellethomas]

lgtm