Add boolean fields

[boraarslan]

The first draft to add boolean field support. There is a very high chance I missed some stuff.

I will update the docs after the review (in case of something needs to change)

[codecov-commenter]

Codecov Report

Merging #1382 (c979d84) into main (5e599d9) will decrease coverage by 0.07%.
The diff coverage is 83.00%.

❗ Current head c979d84 differs from pull request most recent head 9c01d42. Consider uploading reports for the commit 9c01d42 to get more accurate results

@@            Coverage Diff             @@
##             main    #1382      +/-   ##
==========================================
- Coverage   94.26%   94.19%   -0.08%     
==========================================
  Files         235      235              
  Lines       42896    43186     +290     
==========================================
+ Hits        40437    40680     +243     
- Misses       2459     2506      +47     

Impacted Files	Coverage Δ
src/postings/per_field_postings_writer.rs	94.11% <ø> (ø)
src/schema/document.rs	94.11% <0.00%> (-1.89%)	⬇️
src/schema/flags.rs	56.25% <ø> (ø)
src/schema/value.rs	57.08% <21.05%> (-3.20%)	⬇️
src/schema/field_type.rs	80.57% <44.00%> (-3.39%)	⬇️
src/indexer/segment_writer.rs	95.69% <63.63%> (-0.65%)	⬇️
src/schema/term.rs	84.49% <85.00%> (+0.04%)	⬆️
src/query/query_parser/query_parser.rs	94.87% <92.10%> (-0.10%)	⬇️
src/fastfield/mod.rs	94.52% <92.91%> (-0.18%)	⬇️
src/fastfield/writer.rs	90.17% <100.00%> (+0.04%)	⬆️
... and 6 more

---

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 5e599d9...9c01d42. Read the comment docs.

[PSeitz]

Nice job. Checkout test_operation_strategy and extend it to cover boolean fields. It covers different scenarios with randomized input.

[fulmicoton]

@PSeitz @boraarslan Should we merge?

@PSeitz Do we have a mechanism to use the field type to change the fastfield codec logic? It would be nice to force a bitset codec actually... Maybe even drop the DynamicReader for booleans.

[PSeitz]

@PSeitz Do we have a mechanism to use the field type to change the fastfield codec logic? It would be nice to force a bitset codec actually... Maybe even drop the DynamicReader for booleans.

Callingnew_u64_fast_field instead create_auto_detect_u64_fast_field_with_idx always uses Bitpacked encoding. (I think we need to rename that)
But if we have large blocks of true or false, the codec estimator could offer better compression.

[fulmicoton]

@boraarslan @PSeitz Any update?

[PSeitz]

LGTM, setting the codec can imo be done in another PR

[boraarslan]

I agree with @PSeitz but If codec change is a must have I can look into that in this PR.

[fulmicoton]

@PSeitz Can you have approve & merge the PR?