Add polars-bio to the benchmarked tools

[riasc]

Add polars-bio as a benchmarkable interval tool.

What it is: a Python genomics library built on polars / Apache Arrow / DataFusion, offering genomic range operations (overlap/intersect) via a DataFrame API. Installed with pip install polars-bio; overlap is exposed as polars_bio.overlap(df1, df2, ...).

Integration points (follow the existing intervaltree / awk pattern — a small CLI wrapper invoked through tool_call):

• Add a wrapper segmeter/tools/intersect_polars_bio.py that takes -t target -q query -o output and writes BED-format overlaps, mirroring segmeter/tools/intersect_intervaltree.py.
• Add a polars_bio branch in query_call that runs the wrapper via tool_call. segmeter/calls.py (near the intervaltree branch, ~line 270)
• Add "polars_bio" to the --tool argparse choices. segmeter/main.py:35-37
• Decide if a sorted/indexed variant is worth benchmarking; if so add polars_bio_sorted to idx_based_tools and an index_call branch. segmeter/benchmark.py:24-27, segmeter/calls.py:41
• Add it to the Docker image / environment so the benchmark can find it.
• Verify output matches the -wa-style format the precision check expects (first 3 cols = interval), so TP/FP/FN scoring stays comparable.