P9 (WIP): compiled kernels (rust/simurg-core, PyO3) with bit-identical pure-Python fallback

Rust ports of the hot paths found by profiling: fused per-locus GenCall calling
and scoring, BAF/LRR, libm maps with CPython math semantics, both nearest-neighbour
searches of the normalization fit, and the BPM LocusEntry parser. Address
coverage is memoised per IDAT layout. SIMURG_PURE_PYTHON=1 forces the pure paths.
On the Pi: genotype_sample 3.4 s -> 1.1 s, 36-sample GSA run 131 s -> 46 s, BPM
read 4x faster; 36 GSA + 6 Omni2.5 GTCs and QC tables byte-identical to the pure
path; tests/unit/test_core_kernels.py checks every kernel on edge cases.
bench/profile_genotype.py, bench/scaling.py, cloud task p9 (thread scaling).

src/simurg/formats/bpm.py

@@ -32,6 +32,7 @@ from typing import Any
 import numpy as np
 import numpy.typing as npt
 
+from simurg._core import core
 from simurg.core.errors import FormatError
 from simurg.formats.binutil import BinaryReader, load_bytes
 
@@ -223,7 +224,9 @@ def read_bpm(path: str | Path) -> Manifest:
 
 
 def _parse_locus_entries(r: BinaryReader, n: int, name_to_idx: dict[str, int]) -> dict[str, Any]:
-    """Hot loop: parse ``n`` LocusEntry records (optimised pure Python; Rust in phase P9)."""
+    """Hot loop: parse ``n`` LocusEntry records (compiled in ``simurg_core`` when installed; same results)."""
+    if core is not None:
+        return _parse_locus_entries_core(r, n, name_to_idx)
     view = memoryview(r._buf)
     size = r.size
     pos = r.pos
@@ -416,6 +419,53 @@ def _parse_locus_entries(r: BinaryReader, n: int, name_to_idx: dict[str, int]) -
     }
 
 
+_STR_COLS = (
+    "ilmn_ids",
+    "snps",
+    "chroms",
+    "ilmn_strands",
+    "source_strands",
+    "ref_strands",
+    "ploidies",
+    "species",
+    "genome_builds",
+    "sources",
+    "source_versions",
+    "customer_strands",
+    "source_seqs",
+    "top_genomic_seqs",
+    "allele_a_probe_seqs",
+    "allele_b_probe_seqs",
+)
+
+
+def _parse_locus_entries_core(r: BinaryReader, n: int, name_to_idx: dict[str, int]) -> dict[str, Any]:
+    """``_parse_locus_entries`` through ``simurg_core.bpm_locus_entries`` (identical columns and error messages)."""
+    assert core is not None
+    pos = r.pos
+    try:
+        lists, map_infos, a, b, assay, exp_cl, int_only, versions, fracs, used = core.bpm_locus_entries(
+            bytes(r._view[pos:]), n, list(name_to_idx)
+        )
+    except ValueError as e:
+        raise r.fail(str(e)) from None
+    out: dict[str, Any] = dict(zip(_STR_COLS, lists, strict=True))
+    for i in np.flatnonzero(assay < 0):
+        assay[i] = infer_assay_type(out["allele_a_probe_seqs"][i], out["allele_b_probe_seqs"][i], out["source_seqs"][i])
+    r.pos = pos + used
+    return {
+        **out,
+        "map_infos": map_infos,
+        "address_a": a,
+        "address_b": b,
+        "assay_types": assay.astype(np.uint8),
+        "exp_clusters": exp_cl,
+        "intensity_only": int_only,
+        "entry_versions": versions,
+        "base_fractions": fracs,
+    }
+
+
 _IUPAC = {
     "A": 1,
     "C": 2,