Fix buffered MinHashLSH query aggregation across storage backends

[dipeshbabu]

Summary

Fix MinHashLSH.collect_query_buffer() so buffered queries aggregate candidates the same way as repeated calls to query(), including when
using the Cassandra storage backend.

Problem

The buffered query path was intersecting per-band result sets directly. That is stricter than normal LSH query behavior, which unions
candidates across bands for a query and only then intersects across multiple buffered queries.

This caused valid candidates to be dropped when using buffered queries.

The Cassandra backend also exposed a related issue in buffered selects: repeated buffered lookups with the same hash key could be collapsed
instead of preserving one result list per buffered query. That breaks per-query aggregation logic.

Fix

• union bucket hits across bands for each buffered query
• intersect only the per-query candidate sets across the buffer
• preserve existing prepickle behavior
• make Cassandra buffered selects preserve query order and count, including duplicate hash-key lookups
• replace a broken LSH Forest documentation link with a stable reference

Test

• add a regression test showing collect_query_buffer() returns the same candidates as query() for a case where the old implementation
  dropped a valid match

Verification

• confirmed with a direct local repro that buffered and non-buffered query paths now both return [0, 1]
• ran uvx ruff check .
• ran the README test command uv run pytest in Linux/WSL: 158 passed, 76 skipped
• verified the docs link check fix locally with lychee

[gemini-code-assist]

Code Review

This pull request refactors the collect_query_buffer method in datasketch/lsh.py to correctly process buffered queries by unioning candidates across bands for each query before intersecting across the buffer. It also adds a test case to verify that buffered results match direct query results. Feedback identifies a potential bug where the use of zip could truncate results when using the Cassandra storage backend and suggests a more efficient implementation for the set().union() call.

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

❌ Patch coverage is 88.88889% with 1 line in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (master@cc635ed). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
datasketch/lsh.py	87.50%	1 Missing ⚠️
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff            @@
##             master     #307   +/-   ##
=========================================
  Coverage          ?   77.52%           
=========================================
  Files             ?       15           
  Lines             ?     2060           
  Branches          ?        0           
=========================================
  Hits              ?     1597           
  Misses            ?      463           
  Partials          ?        0           

Flag	Coverage Δ
unittests	77.52% <88.88%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.