bugfix for duckdb engine, multi-engine support, perf regression investigate

benchmark/h2o/groupby.sql

@@ -0,0 +1,22 @@
+SELECT id1, sum(v1) AS v1 FROM source GROUP BY id1; --0.036 |  0.017 | .274
+
+SELECT id1, id2, sum(v1) AS v1 FROM source GROUP BY id1, id2; -- 0.063 | 0.013
+
+SELECT id3, sum(v1) AS v1, avg(v3) AS v3 FROM source GROUP BY id3; -- 2.322 | 0.406 | 2.27
+
+SELECT id4, avg(v1) AS v1, avg(v2) AS v2, avg(v3) AS v3 FROM source GROUP BY id4; -- 0.159 | 0.022
+
+SELECT id6, sum(v1) AS v1, sum(v2) AS v2, sum(v3) AS v3 FROM source GROUP BY id6; -- 1.778 | 0.699 | 2.283
+
+--faster median
+--SELECT id4, id5, median(v3) AS median_v3, stddev(v3) AS sd_v3 FROM source GROUP BY id4, id5; -- x4
+
+SELECT id3, max(v1) - min(v2) AS range_v1_v2 FROM source GROUP BY id3; -- 0.857 |  0.467 | 2.236
+
+-- select top 2 from each grp
+-- SELECT id6, subvec(v3,0,2) AS v3 FROM source GROUP BY id6 order by v3;
+
+-- implement corr
+--SELECT id2, id4, pow(corr(v1, v2), 2) AS r2 FROM source GROUP BY id2, id4;
+
+SELECT id1, id2, id3, id4, id5, id6, sum(v3) AS v3, count(*) AS cnt FROM source GROUP BY id1, id2, id3, id4, id5, id6; -- 2.669 | 1.232 | 2.221(1.8)

benchmark/h2o/groupby_prep.sql

@@ -0,0 +1,6 @@
+create table source(id1 int,id2 int,id3 int,id4 int,id5 int,id6 int,v1 int,v2 int,v3 float)
+
+
+LOAD DATA INFILE "data/h2o/G1_1e7_1e1_0_0_n.csv"
+INTO TABLE source
+FIELDS TERMINATED BY ","