Ensure that BitVec has a deterministic size

[aneubeck]

There was some "unexpected" behaviour in BitVec.
It took as argument a size, but then replaced it (sometimes) with the larger bit position?!?
I removed that hack and fixed the calling places.

[Copilot]

Pull Request Overview

This PR fixes deterministic sizing behavior in BitVec by removing logic that would sometimes replace the intended size with the largest bit position. The changes ensure BitVec maintains its specified size consistently and updates all calling code to work with this corrected behavior.

Key changes:

• Removed conditional size replacement logic in BitVec::from_bit_chunks that created non-deterministic behavior
• Fixed GeoDiffCount construction and bit manipulation logic to handle the corrected BitVec sizing
• Introduced a test harness for deterministic random number generation across all tests

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
crates/geo_filters/src/test_rng.rs	Adds new test harness for seeded RNG with panic recovery
crates/geo_filters/src/lib.rs	Imports the new test_rng module for test builds
crates/geo_filters/src/distinct_count.rs	Updates tests to use new RNG harness
crates/geo_filters/src/diff_count/bitvec.rs	Removes size replacement logic and unused parameter
crates/geo_filters/src/diff_count.rs	Fixes bit manipulation logic and updates tests
crates/geo_filters/src/config/lookup.rs	Updates tests to use new RNG harness
crates/geo_filters/src/config.rs	Updates test estimation function to use new RNG harness

Comments suppressed due to low confidence (1)

crates/geo_filters/src/diff_count.rs:523

• [nitpick] Running the same test 1000 times with different random seeds may be excessive and could significantly slow down test execution. Consider reducing this to a more reasonable number like 10-100 iterations.

        for _ in 0..1000 {

[Copilot]

The chunks parameter is no longer used after removing the peek operation. Consider removing the Peekable wrapper and use I directly since the parameter is only passed to iter_ones which can handle a regular iterator.

Suggested change

	chunks: Peekable<I>,
	chunks: I,

[itsibitzi]

Following a discussion on Slack with Alex I've cherry-picked these changes onto my branch. Going to close this PR.