LowCardinality Dictionary as skip index

[UnamedRus]

Use case

Clickhouse 20.8 
CREATE TABLE test_lc (partition UInt32, key UInt32, value LowCardinality(String)) ENGINE=MergeTree PARTITION BY partition ORDER BY key;
INSERT INTO test_lc SELECT 0 as a, number, 'dasdsadsdsds' FROM numbers(100000000);
INSERT INTO test_lc SELECT 1 as a, number, 'dasdfsagfgdgs' FROM numbers(100000000);
INSERT INTO test_lc SELECT 2 as a, number, 'sadasdasdasdas' FROM numbers(100000000);
INSERT INTO test_lc SELECT 2 as a, number, 'special value' FROM numbers(11);
SELECT * FROM test_lc WHERE value = 'special value';
2020.11.05 15:34:28.897084 [ 5013 ] {9448e6c4-916c-40bd-96d2-dad4c7273372} <Information> executeQuery: Read 300000011 rows, 286.28 MiB in 0.2246097 sec., 1335650290 rows/sec., 1.24 GiB/sec.
2020.11.05 15:34:28.897180 [ 5013 ] {9448e6c4-916c-40bd-96d2-dad4c7273372} <Debug> MemoryTracker: Peak memory usage (for query): 0.00 B.

11 rows in set. Elapsed: 0.226 sec. Processed 300.00 million rows, 300.19 MB (1.32 billion rows/s., 1.33 GB/s.)

Describe the solution you'd like
Use lowCardinality dictionary to filter out parts which don't have value for which we are searching.
In case of multiple lowCardinality dictionaries for a single part, it can filter out some ranges from certain parts.

Describe alternatives you've considered
Create secondary indices to handle that kind of queries.

[alexey-milovidov]

PREWHERE works very similarly, although reading only the dictionary part will be better.