[rowexec] custom rowexec

[max-hoffman]

This PR adds custom Dolt execution operators for lookup joins. When building an execution plan, we try to replace joinIter with a Dolt equivalent that inlines the key building and map get. This is a lot faster than repeatedly building the secondary iterator and materializing sql.Rows in-between lookups.

The main downside is that this PR hoists filters in join children to after materializing lookup join rows.

This brings index_join from 5.18 ms/query to 2.64 ms/q, which will be about 2.0x MySQL's latency.

This PR falls short of some aspiration goals:

• We hoist table filters until after the final join row is built because we don't have a way to call scalar expressions on val.Tuple yet. There are edge case queries that might be dramatically slower because of this. To fix this, we could need to convert sql.Expression filters into a format that we could execute on val.Tuple KV pairs.
• We do not yet try to to optimize consecutive lookup joins. I'm not sure if a materialization block would be better represented iteratively or recursively beyond a simple string of lookups. A lot of interfaces and indexing considerations to think about there.

Safety comments:

• we fallback to GMS when lookup source/dest keys are not prolly.Encoding compatible
• the source iterators are the same as what we used before, but without projection mapping to sql.Rows. The keyless iterator required a change to return duplicate rows at the KV layer (vs the sql layer).
• the secondary iterators are a generalization of what we currently use, but return KV pairs instead of rows
• projection mapping is the same but generalized to merge an arbitrary list of KV pairs after the join

There are extra tests here: dolthub/go-mysql-server#2593

[max-hoffman]

#benchmark

[github-actions]

@max-hoffman workflow run: https://github.com/dolthub/dolt/actions/runs/9682864996

[coffeegoddd]

@max-hoffman DOLT

test_name	from_latency_p95	to_latency_p95	is_faster
tpcc-scale-factor-1	73.13	89.16	0

test_name	server_name	server_version	tps	test_name	server_name	server_version	tps	is_faster
tpcc-scale-factor-1	dolt	f7abadf	32.99	tpcc-scale-factor-1	dolt	c531bce	13.52	1

[coffeegoddd]

@max-hoffman DOLT

read_tests	from_latency_median	to_latency_median	is_faster
covering_index_scan	2.76	2.81	0
groupby_scan	17.32	17.32	0
index_join	5.28	2.61	1
index_join_scan	2.57	2.57	0
index_scan	53.85	54.83	0
oltp_point_select	0.46	0.46	0
oltp_read_only	7.56	7.7	0
select_random_points	0.77	0.77	0
select_random_ranges	0.92	0.92	0
table_scan	54.83	55.82	0
types_table_scan	142.39	142.39	0

write_tests	from_latency_median	to_latency_median	is_faster
oltp_delete_insert	6.09	6.09	0
oltp_insert	3.02	3.02	0
oltp_read_write	14.21	14.21	0
oltp_update_index	3.13	3.07	0
oltp_update_non_index	3.02	3.02	0
oltp_write_only	6.43	6.43	0
types_delete_insert	6.67	6.67	0

[github-actions]

Additional work is required for integration with DoltgreSQL.

[max-hoffman]

#benchmark

[github-actions]

@max-hoffman workflow run: https://github.com/dolthub/dolt/actions/runs/9687941574

[coffeegoddd]

@max-hoffman DOLT

test_name	from_latency_p95	to_latency_p95	is_faster
tpcc-scale-factor-1	74.46	78.6	0

test_name	server_name	server_version	tps	test_name	server_name	server_version	tps	is_faster
tpcc-scale-factor-1	dolt	0c4b3f9	32.94	tpcc-scale-factor-1	dolt	ef8cc8b	32.89	0

[coffeegoddd]

@max-hoffman DOLT

read_tests	from_latency_median	to_latency_median	is_faster
covering_index_scan	2.86	2.86	0
groupby_scan	17.01	17.32	0
index_join	5.28	2.66	1
index_join_scan	2.52	2.57	0
index_scan	53.85	53.85	0
oltp_point_select	0.44	0.46	0
oltp_read_only	7.43	7.56	0
select_random_points	0.73	0.74	0
select_random_ranges	0.87	0.89	0
table_scan	54.83	54.83	0
types_table_scan	139.85	142.39	0

write_tests	from_latency_median	to_latency_median	is_faster
oltp_delete_insert	6.09	6.09	0
oltp_insert	2.97	3.02	0
oltp_read_write	13.7	13.95	0
oltp_update_index	3.07	3.07	0
oltp_update_non_index	3.02	3.02	0
oltp_write_only	6.32	6.43	0
types_delete_insert	6.55	6.67	0

[coffeegoddd]

@max-hoffman DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
c50e5b8	ok	5937457

version	total_tests
c50e5b8	5937457

correctness_percentage
100.0

[coffeegoddd]

@coffeegoddd DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
921e758	ok	5937457

version	total_tests
921e758	5937457

correctness_percentage
100.0

[max-hoffman]

#benchmark

[github-actions]

@max-hoffman workflow run: https://github.com/dolthub/dolt/actions/runs/9880357397

[coffeegoddd]

@max-hoffman DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
ef3a0cf	ok	5937457

version	total_tests
ef3a0cf	5937457

correctness_percentage
100.0

[coffeegoddd]

@max-hoffman DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
f5ab6be	ok	5937457

version	total_tests
f5ab6be	5937457

correctness_percentage
100.0

[coffeegoddd]

@coffeegoddd DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
327ce5e	ok	5937457

version	total_tests
327ce5e	5937457

correctness_percentage
100.0

[max-hoffman]

#benchmark

[github-actions]

@max-hoffman workflow run: https://github.com/dolthub/dolt/actions/runs/9912163940

[coffeegoddd]

@max-hoffman DOLT

test_name	from_latency_p95	to_latency_p95	is_faster
tpcc-scale-factor-1	74.46	77.19	0

test_name	server_name	server_version	tps	test_name	server_name	server_version	tps	is_faster
tpcc-scale-factor-1	dolt	3745baa	32.69	tpcc-scale-factor-1	dolt	327ce5e	32.51	0

[coffeegoddd]

@max-hoffman DOLT

read_tests	from_latency_median	to_latency_median	is_faster
covering_index_scan	3.02	3.02	0
groupby_scan	17.32	17.32	0
index_join	5.28	5.28	0
index_join_scan	2.61	2.61	0
index_scan	55.82	55.82	0
oltp_point_select	0.46	0.46	0
oltp_read_only	7.84	7.84	0
select_random_points	0.75	0.75	0
select_random_ranges	0.9	0.9	0
table_scan	56.84	57.87	0
types_table_scan	147.61	147.61	0

write_tests	from_latency_median	to_latency_median	is_faster
oltp_delete_insert	6.09	6.09	0
oltp_insert	3.02	3.02	0
oltp_read_write	14.21	14.21	0
oltp_update_index	3.13	3.13	0
oltp_update_non_index	3.02	3.02	0
oltp_write_only	6.43	6.43	0
types_delete_insert	6.67	6.67	0

[max-hoffman]

Trying to figure out how to add a unit test for making sure we do the optimization when we expect. Surprisingly easy to accidentally disable.

[coffeegoddd]

@max-hoffman DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
ba424ee	ok	5937457

version	total_tests
ba424ee	5937457

correctness_percentage
100.0

[max-hoffman]

#benchmark

[github-actions]

@max-hoffman workflow run: https://github.com/dolthub/dolt/actions/runs/9914037567

[coffeegoddd]

@max-hoffman DOLT

test_name	from_latency_p95	to_latency_p95	is_faster
tpcc-scale-factor-1	75.82	75.82	0

test_name	server_name	server_version	tps	test_name	server_name	server_version	tps	is_faster
tpcc-scale-factor-1	dolt	3745baa	32.33	tpcc-scale-factor-1	dolt	ba424ee	32.28	0

[coffeegoddd]

@max-hoffman DOLT

read_tests	from_latency_median	to_latency_median	is_faster
covering_index_scan	3.02	3.02	0
groupby_scan	17.32	17.32	0
index_join	5.28	2.71	1
index_join_scan	2.61	2.61	0
index_scan	54.83	54.83	0
oltp_point_select	0.46	0.46	0
oltp_read_only	7.7	7.7	0
select_random_points	0.75	0.75	0
select_random_ranges	0.9	0.9	0
table_scan	56.84	56.84	0
types_table_scan	144.97	144.97	0

write_tests	from_latency_median	to_latency_median	is_faster
oltp_delete_insert	5.99	5.99	0
oltp_insert	2.97	3.02	0
oltp_read_write	13.95	13.95	0
oltp_update_index	3.07	3.07	0
oltp_update_non_index	2.97	3.02	0
oltp_write_only	6.32	6.43	0
types_delete_insert	6.55	6.67	0

[coffeegoddd]

@max-hoffman DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
8a59480	ok	5937457

version	total_tests
8a59480	5937457

correctness_percentage
100.0

[coffeegoddd]

@coffeegoddd DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
bf6fdbb	ok	5937457

version	total_tests
bf6fdbb	5937457

correctness_percentage
100.0

[coffeegoddd]

@max-hoffman DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
425b421	ok	5937457

version	total_tests
425b421	5937457

correctness_percentage
100.0

[coffeegoddd]

@coffeegoddd DOLT

comparing_percentages
100.000000 to 100.000000

version	result	total
cc67078	ok	5937457

version	total_tests
cc67078	5937457

correctness_percentage
100.0