Add LD-pruned biallelic SNP access method

[adilraza99]

Summary

Add support for generating LD-pruned SNP datasets from the existing biallelic SNP call pipeline.

Changes

• introduce AnophelesLdAnalysis mixin
• add biallelic_snp_calls_ld_pruned() method
• add ld_params.py for LD pruning parameters
• integrate the mixin into AnophelesDataResource
• add tests for LD pruning behaviour

Implementation

The method builds on top of biallelic_snp_calls():

1. load SNP dataset using the existing pipeline
2. compute genotype reference counts
3. apply LD pruning using allel.locate_unlinked
4. filter variants using _dask_compress_dataset
5. return an LD-pruned xarray.Dataset

Notes

• implementation follows the repository's cooperative mixin architecture
• returned dataset remains compatible with downstream processing (including PLINK export)
• LD pruning operates on datasets optionally thinned using the existing n_snps parameter

Tests

Added tests to verify:

• variant count is reduced after pruning
• dataset structure remains valid
• compatibility with downstream workflows

---

closes #1049

[adilraza99]

Hi @jonbrenas,

I ran a quick local check of the new method on a small region. The LD-pruned dataset returns fewer variants than the original SNP calls and behaves as expected with the existing pipeline.

Attaching a screenshot from the local run for reference.

[adilraza99]

@jonbrenas could you take a look at this when you have a moment?

[adilraza99]

Hi @jonbrenas, thanks for the clarification. I tried an implementation based on the discussion and also ran a quick local verification where the variant count decreases after pruning. When you have time, could you please take a look and let me know if this direction makes sense?