Bulk decompression of text columns

[akuzm]

Implement bulk decompression for text columns. This will allow us to use them in the vectorized computation pipeline.

No changes in performance: https://grafana.ops.savannah-dev.timescale.com/d/fasYic_4z/compare-akuzm?orgId=1&var-branch=All&var-run1=3093&var-run2=3078&var-threshold=0&var-use_historical_thresholds=true&var-threshold_expression=2.5%20*%20percentile_cont(0.90)&var-exact_suite_version=false

Disable-check: force-changelog-file

[codecov]

Codecov Report

Attention: 44 lines in your changes are missing coverage. Please review.

Comparison is base (85b27b4) 79.79% compared to head (7985c76) 79.78%.

Files	Patch %	Lines
tsl/src/compression/array.c	76.27%	0 Missing and 14 partials ⚠️
tsl/src/compression/dictionary.c	76.36%	0 Missing and 13 partials ⚠️
tsl/test/src/decompress_text_test_impl.c	71.42%	4 Missing and 6 partials ⚠️
tsl/src/nodes/decompress_chunk/compressed_batch.c	86.27%	3 Missing and 4 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #6458      +/-   ##
==========================================
- Coverage   79.79%   79.78%   -0.01%     
==========================================
  Files         190      190              
  Lines       36949    37101     +152     
  Branches     9356     9403      +47     
==========================================
+ Hits        29482    29600     +118     
- Misses       3116     3127      +11     
- Partials     4351     4374      +23     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[github-actions]

@fabriziomello, @antekresic: please review this pull request.

Powered by pull-review

[svenklemm]

Can we have test files/functions in dedicated direcotry?

[antekresic]

Is there a reason why the PR is light on sql tests? I see a bunch of fuzzing tests were added but that's mostly it.

[akuzm]

Is there a reason why the PR is light on sql tests? I see a bunch of fuzzing tests were added but that's mostly it.

Since the bulk decompression is enabled by default, it is covered by many existing tests that use text columns. I plan to add some SQL tests when we have vectorized filters on text columns: https://github.com/timescale/timescaledb/pull/6189/files#diff-0068ab49636b29bd1b20ebdd4f87f04251dae1a461f499500179100eab2c2494

[akuzm]

Can we have test files/functions in dedicated direcotry?

Moved them to tsl/test/src. I also tried to make a proper CMake flag for fuzzer support, that would also find the libraries and configure the compiler, but I can't make the CMake work. That directory is a separate library, but it's not actually linked as library but is somehow concatenated to tsl library using TARGET_OBJECTS... So for now the compiler flags and library paths are still specified manually in the libfuzzer workflow.