perf: precompute address hash instead of per-instance lru_cache - #1774

What

Replace the per-instance functools.cache(lambda ...) assigned to self.__hash__ in ZeroconfIPv4Address / ZeroconfIPv6Address with a hash precomputed once into an int slot, exposed through a real __hash__ method.

Why

Same class of bug as Bluetooth-Devices/cached-ipaddress#108. The old code did:

self.__hash__ = cache(lambda: IPv4Address.__hash__(self))

Because __hash__ was listed in __slots__, the memoization did take effect here (unlike the cached-ipaddress case, where it was dead code). But every address object still paid to allocate a full lru_cache wrapper and a self-capturing lambda at construction, plus created a self → slot → wrapper → lambda → self reference cycle the GC has to collect. get_ip_address_object_from_record builds one of these per A/AAAA record on a cache miss, so the construction cost is on the discovery hot path.

How

• Compute self._hash = IPv4Address.__hash__(self) (resp. IPv6) once in __init__, store in a plain int slot.
• Add a real __hash__ method returning it — defined on the type, so dunder lookup resolves correctly and no per-instance wrapper is allocated.
• Swap "__hash__" → "_hash" in __slots__; drop the now-unused functools.cache import and the type: ignore[method-assign].

No cdef class declaration for these types in ipaddress.pxd, so they stay pure-Python classes under the Cython build; cython -3 compiles the module cleanly.

Testing

• SKIP_CYTHON=1 poetry run pytest tests/utils/test_ipaddress.py tests/services/test_info.py tests/test_listener.py tests/utils/ — 85 passed, 1 skipped, 1 xpassed.
• New regression test pins hash equality with stdlib addresses and set-dedup of equal addresses.
• Pure-Python microbenchmark: IPv4 construction ~14.7µs → ~10.8µs per object; hash stays memoized.

Stacked on the benchmark PR #1773 — this branch includes that commit so CodSpeed can measure the delta. Once #1773 merges, rebasing on master narrows this PR to the fix commit alone.

Quality Report

Changes: 3 files changed, 75 insertions(+), 5 deletions(-)

Code scan: clean

Tests: passed (4 PASSED)

Branch hygiene: clean

Generated by Kōan post-mission quality pipeline