chore: update semantic conventions to v1.44.0 (#5511)

* fix(semconv): filter duplicate metric refinements in generator template

Weaver v0.24.2 emits one metric entry per refinement (e.g. hw.errors is
refined per hardware component: GPU, disk, network adapter, etc.), all
sharing the same metric_name. The Python semantic_metrics.j2 filter
passed all of them through unfiltered, producing duplicate
create_hw_errors/create_hw_status definitions.

Mirrors the workaround applied in semantic-conventions-java#549: keep
only the entry whose id matches its own canonical metric_name, dropping
the per-component refinement duplicates.

A real fix in weaver itself is tracked at
https://github.com/open-telemetry/weaver/pull/1676.

Assisted-by: Claude Sonnet 4.6

* chore: update semantic conventions to v1.44.0

Bump SEMCONV_VERSION in scripts/semconv/generate.sh from 1.43.0 to
1.44.0, add V1_44_0 to schemas.py, and regenerate all attribute and
metric constants with weaver v0.24.2.

Notable additions in v1.44.0 (messaging / Kafka):
- messaging.kafka.cluster.id (incubating)
- messaging.destination.template (incubating)
- messaging.message.conversation_id (incubating)
- messaging.message.envelope.size (incubating)

Assisted-by: Claude Sonnet 4.6

* chore: add changelog fragment for semconv v1.44.0 update

Assisted-by: Claude Sonnet 4.6